AMC3306M25: diag pin triggering

Hi Steve,

Is it possible to share the layout? Gerbers would be fine. If you are uncomfortable sending publicly, you can always send a friend request and you can share it privately. It seems unlikely that both devices would be faulty and I would first like to determine if there could be any potential causes. However, just as a sanity check, you can replace on the 'bad' devices with a known good one to see if the behavior persists. 

Additionally, have you probed the DCDC out pin? If you can add a probe so we can monitor the status of the DCDC out when the DIAG pin goes low, that would be helpful. 

Regards,
Aaron Estrada

Hi Aaron,

Ok. so carried out the measurements as requested. N.B my scope and differential probe not that accurate so we should compare the relative differences in the traces:

Faulty lane 1 - DCDC out on green

Faulty lane 2 - DCDC out on green

healthy lane 3 - DCDC out on green

using a DVM the following readings were taken:

lane 1 HLDOout - 3.2009 DCDCout - 3.49

lane 2 HLDOout - 3.2129 DCDCout - 3.49

lane 3 HLDOout - 3.2054 DCDCout - 3.527

So there is a difference between the healthy lane and the faulty lanes.

PCB layouts:

The local layout to each chip is pretty much identical. The working one is the one farthest away from the ribbon connector.

There is a ribbon cable (10cm long) between this power interface PCB and the control PCB at the control PCB end the following circuit is driving the interface:

Note the local power for the external side of the SDFM interface is from a RECOM 15V/3V3 DC/DC converter.

I note on your evaluation board you have an additional 10uF capacitor on the 3V3 supply and ours is local to the 15V/3V3 DC/DC converter. Should we consider beefing up the capacitance at the remote end?

Any other ideas?

I note these chips are out of stock at the moment so we are going to be a bit stuck if the conclusion is that we have two faulty ones.

regards

Steve

Aaron,

Ok so I just moved my scope ground from TP23 on the control board to a local ground on the power interface as I was going to take a look at the remote 3V3 to the devices. All three sensors seem have started working the same now.  A rather unexpected finding, but worth reporting as it may shed some light on the issue...

However, if I remove the scope entirely, the indications from the microprocessor is suggesting all three are still working ok.

Now, I just put the scope ground back on TP23 and it has continued to work on all three channels.  So this may be a red herring or perhaps something is temperature sensitive or..... 

Dropped my clock frequency back to 5MHz and it is still stable on all three....

Any thoughts?

regards

steve

Hi Steve,

This is very interesting. So to confirm, you have everything working as intended? Regarding the bypass caps on the 3.3V supply, I would like to confirm if you have capacitors close to the VDD pin of the AMC. If they are only on the source side, then that may be a little concerning. We recommend that bypass capacitors be placed as close to the device pins as possible. 

I will review the information you presented in greater detail and get back to you shortly. 

Regards,

Aaron Estrada

Hi Aaron,

All the decoupling caps are close to the devices as per the reccomendations in the datasheet. Yes it started running ok after an hour or so and continued to do so for several hours.

I guess if i start it up again tomorrow and it fails initially we might suspect temperature or some such.

Regards

Steve

Hi Steve, 

That is interesting. Perhaps there could have been a grounding issue and when you initially added the probe, it somehow shifted something the right way and resolved it. Taking a deeper look, I would also like to know where you got the parts from. Do you have the name of the manufacturer? If you can send over the markings on the top of the IC for all 3 parts, that would be great. We can also take a look to check some information on the devices. 

Additionally, I am trying to understand how things are routed on the PCB of the photos you sent. Do you have a full schematic for the board with the AMC's on it? I am trying to see what these diodes and large caps are used for. 

Regards,

Aaron Estrada

Aaron,

the overall circuit is just a three-phase bridge with the three devices as previously provided. They are connected to the main control board via the header shown:

We have two of these boards built and I briefly ran both yesterday for a few minutes and both seemed to be working ok. (The data in the microprocessor was stable for all three channels on both boards).  I probably need to do a longer test to satisfy myself they are both operating without any glitches.

I can't quite imagine why attaching a scope ground to the supply ground should make a difference to two of the devices and not the third, that's very odd....

The devices were all bought from Mouser I believe, probably about a year ago when they still could be found in the world...