DRV8873HEVM: Jitter on DRV8873H outputs in PWM mode

Thanks Ryan - I had thought that there would be some jitter of that magnitude on the outputs due to input re-synchronisation to the 10MHz internal clock, but we are seeing anywhere from 400ns to 800 ns depending on how you measure it so there may be something else going on.

It's a bit late for our project to re-design the boards, and looking at the DRV8801A although it looks like it can handle the current it lacks some attractive features of the 8873 such as programmable slew rate.  It also has quite a high on-resistance which would cause us junction temperature problems due to the much higher dissipation than the 8873 (Rds(on) = 0.155ohm).

Would you not expect to see the level of jitter we're seeing even on the eval board?

I tried to upload a Picoscope file but this interface doesn't seem to allow it.

BR,
Ian

Ian,

I am out of the office now, but when I get back on Tuesday, July 6, I will check some data.  Will get back to you then.

Regards,

Ryan

Thanks Ryan.

I've looked again at the DRV8801 and I don't think it will solve our problem - the Rds(on) is just too high and the stated package derating of 27mW/C above 25C doesn't help us.

What I would like to know if possible is:
- is the jitter that we are observing both on our boards and on the EVM higher than would be expected (400-800ns)?
- if so, is there any way to reduce it?
- if not, then does TI have a list of all H-bridge devices that don't synchronise their inputs and so should not exhibit output jitter

Our problem is that we now have a fairly wide dead band around 0A in our current control loop because of this jitter.  We have tried increasing the slew rate and this helps a little (but is not significantly better), and we might run into emissions problems which we won't know until we carry out our EMC testing.

Although we only have a load current of 1A in this design, we have other applications with higher load currents so understanding the range of devices available is important.

Our applications are automotive with a high ambient temperature, so a low Rds(on) will be critical.

BR,
Ian

Ian,

Please give me 24 hours.  Is automotive qual a must have?  Would open up some other options if not required.

Regards,

Ryan

Ryan - automotive qual is highly preferred but we would look at other options, part availability is really poor at the moment especially auto-qual parts.

ian

Ian,

Conditions may be different from your testing, but this scope capture shows ~200ns for jitter.  Inputs are GREEN and RED in image below.  Ignore the YELLOW trace.  

To answer and earlier question, there is no way to improve this on 8873.  We have an industrial part, DRV8844, that will work without this effect and that is about it for now.  In ~ 1 month, we will prototype release some lower Rdson versions of the DRV8873 with an option to disable the SS clocking feature.  This won't completely eliminate the jitter, but it will improve it.  These devices are not public yet so I can't point you to a datasheet just yet, but depending on your schedule, they might work.  

Regards,

Ryan

Thanks for your help Ryan.

At least we know that we can't do any better with the current device.

I'll look again at the jitter on the EVM to see if I can correlate with your measurement, and we may look at the 8844 as it may do a better job for our low-current applications than the 8873, shame it's not automotive.

By the way - I think this site needs to change the 'resolved/not resolved' flags to 'answered/not answered' as the issue isn't resolved but the question has been answered, so no further support is needed at this time and you can close/lock this issue.

Cheers,
Ian