DRV8840 output jitter

Jonathan,

If I am reading the scope capture right, this is a <=1us pulse you are issuing on the enable.  I think what you are seeing is our deglitch circuit acting on this narrow pulse.

I need to check with design to see what our deglitch timing is on the ENABLE pin as I can't find this information in the datasheet.

If you increase the low time, does the problem go away?

Hi Ryan,

Yes, the scope capture is 1us/div. I was unaware of the deglitch circuit, but it would explain a lot.

Running the EVM software, I see pulse-skipping on the outputs typically at 98 and 99% duty cycle. This is the point where the off-time becomes < 1us. There can be milliseconds between output pulses, hence it puts it in the audible range, and increases the ripple to the TEC as my LC filter is no longer effective.

I was planning on running at 62kHz to keep my filter smaller (the EVM runs at 31kHz). I now see this will make things worse by way of narrower pulses.

The TEC will be driven by PID control, so if it's far from the set-point it will run flat-out until it gets close, then it will pass through this high duty cycle region as it backs off. I suppose I could restrict the maximum duty at the expense of response time and range.

There is also a 300 to 400ns jitter on rising and falling edges of the output pulse (for any duty cycle), which is contributing to audible noise, although nowhere near as badly as the pulse-skipping.

If you know of any alternative parts that may be better suited to the application, please let me know!

Jonathan,

You could consider the DRV8842 which has a direct PWM interface instead of the Phase/Enable interface. 

Other options with direct PWM interface and very fast (~14ns) edges are the DRV8412.  This is currently being used in TEC applications today. 

Hi Guys,

Just thought of mentioning that the DRV8844 should have way less jitter than the DRV8841/43 devices. Although the input/output structure is kind of different (DRV8844 is in essence a quad half H Bridge with tri-state control) you can still use the same layout as the DRV8841/43 devices.

Hi Ryan,

Does the DRV8842 not have the deglitch circuit that's in the DRV8840? Is there a more 'direct' relationship between PWM in and PWM out?

I looked at the DRV8412 earlier, but got the impression it can't do 100% duty - please correct me if I'm wrong.

Hi Avayan,

Thanks for the tip, although I'm curious as to how you know which devices should have less jitter. They all have a mysterious box labelled "control logic" on the datasheet.

Also, with the DRV8844, I'd need to know if I can put two outputs in parallel as I need 2.5A rms. I couldn't see that mentioned on the datasheet.

Hi Jonatahn,

I have been tinkering with DRV88xx devices for a VERY long time! You can read about some of my experiments and findings at www.ebldc.com. BTW, I am using the DRV8844 as my power stage to drive the hot element on my 3D printer. I also use one of its outputs to drive te fan which I use to cool down the plastics.

On some oof my experiments with DRV8844, I noticed the outputs respond much quicker to the inputs, than on the DRV8841/43.

And yes, you should be able to parallel outpus. Just make sure the connection is as strong as possible and as close to the chip as possible.