DRV8412 single bridge current waveform deformation

Hi Sergio,

If you haven't done so, please try the following:
1) Swap the inductive loads from bridge AB to CD. Does the noisy current move to AB? If so, it is the inductive loads. Please look for extra capacitance on the load.
2) Can you provide a schematic and layout? Is the bridge CD layout significantly different than the bridge AB layout?

Hi Rick,

thank you for your quick reply.

Here the answers to your questions:

1) I have already tried to swap the loads but without any difference. The noisy current still come from bridge CD.

I add some more details about the load I am trying to drive:

the load is composed by an inductance about 330uH placed close to the board connector and it is connected by a 1.3m (1300millimeters) cable to the coil (that move a mechanical parts of the machine); the coil has a nominal inductance value about 420uH.

2) about the layout, the two output channels (from bridge AB & CD) are quite similar, as you can see by the attached file.

Here you find the schematic diagram about the first DRV8412. The second has the same schematic.

DRV8412 power stage.pdf

And here the power stage layout:

DRW8412 power stage layout.pdf

Best regards,

Sergio

Hi Sergio,

Thank you for the additional information. We have not found anything that is obvious, so it looks like more investigation is required. It was noted the OUT_D path appears to be different from the other traces.

1) Do you have a TI DRV8412EVM? If so, have you tried running the loads on it? What were the results?
2) Would you capture the voltages in addition to currents at the connectors? Is there any difference between bridge AB and CD?
3) When swapping the good load onto CD, did you also disable the AB side? This may provide an additional clue.
4) Can you confirm that the same current probe was used in both measurements? Although unlikely, we want to rule out equipment.
5) Can you confirm that the current probe is placed on the load wire for both measurements?
6) It appears the trace for OUT_D is different than the other outputs, using multiple layers to get to the connector. You might consider looking at it in more detail. If this is suspect, can you consider modifying a board by cutting the existing path and wiring directly from the pin to the inductor on one board?

Hi Rick,

thank you for your help.

At this moment I can give you an immediate feedback only for some of your questions.

Q1) Do you have a TI DRV8412EVM? If so, have you tried running the loads on it? What were the results?

A1) I haven't tested the load on a 8412 demo. We have already used the DRV8412 on older projects without any problem,

so I skipped this step. Usually we ask demo board only for never used devices.

I think that we can perform some intermediate steps before test the load on a demo but I agree with the idea.

Q2) Would you capture the voltages in addition to currents at the connectors? Is there any difference between bridge AB and CD?

A2) I haven't captured the voltages on the load yet, but I think I can get some measurement soon.

Q3) When swapping the good load onto CD, did you also disable the AB side? This may provide an additional clue.

A3) No, the AB bridge was ON. I add that the two "current commands" in the two loads do not start (and finish) exactly

at the same time, but they can be time shifted, depending on the machine commands.

I will do this test as soon as possible.

Q4) Can you confirm that the same current probe was used in both measurements? Although unlikely, we want to rule out equipment.
Q5) Can you confirm that the current probe is placed on the load wire for both measurements?

A4/A5) Same current probe used, properly degaussed and reset to zero, same scope. Probe connected to the corresponding load wire under test.

Q6) It appears the trace for OUT_D is different than the other outputs, using multiple layers to get to the connector. You might consider looking at it in more detail. If this is suspect, can you consider modifying a board by cutting the existing path and wiring directly from the pin to the inductor on one board?

A6) I strongly agree with your suspect on OUT_D trace. I will do what you suggest ASAP, and I will report the results.

Thanks for all, have a nice day,

Sergio

Hi Rick,

I have some more news to report.

Yesterday I managed to obtain a board to test in laboratory, not on the field.

I set up the board with the same loads, same cable, same power supply and same additional inductors (placed on a little PCB with the same connectrors of the DRV8412 Driver board).

Sadly I have to report that on this board, the ringing current is swapped on the AB bridge, on both the DRV8412; to be clear where we have the "good" layout side, where the PCB traces run always on top layer.

Following your questions, I got this results:

Q2) output voltages

On output connector I saw a square wave (from 0V to +24V) with an overshoot on the "corners" with a maximum amplitude amplitude around 2.2V (see the following picture).

Observing the same voltage signal on the DRV8412 pin I see a reduced overshoot amplitude. I think the ferrite L10 is doing its duty.

Moreover, this also means that the problem comes from outside to the DRV.

Q3) enable one bridge per time

I managed to activate one bridge separately from the other. What I see is that the activation of the second bridge increase the current ringing on the first.

Q6) PCB traces

I cut the connection on the bottom layet on the CD bridge and replaced it with a short pice of large diameter wire; I expected to see the DC bridge current getting whorse, similar to the AB bridge current.

On the contrary, The current shapes on AB bridge (dirty) and CD bridge (clean) stay the same.

After this I concentrated my attention on the cable harness. Here I summarize the results of multiple tests:

1A) machine cable + external inductors (330uH) + AB channel on -> peak to peak current ripple on AB out = 260mA

1B) machine cable + external inductors (330uH) + AB and DC channel on -> peak to peak current ripple on AB out = 340mA (as previously said)

2A) short and separate wire cable + external ind (330uH) + AB channel on -> peak to peak current ripple on AB out = 80mA

2B) short and separate wire cable + external ind (330uH) + AB and DC channel on -> peak to peak current ripple on AB out = 80mA

3A) long and separate wire cable + external ind (330uH) + AB channel on -> peak to peak current ripple on AB out = 118mA

3B) long and separate wire cable + external ind (330uH) + AB and DC channel on -> peak to peak current ripple on AB out = 124mA

this means that the lenght increase the problem

4A) long and grouped wire cable + external ind (330uH) + AB channel on -> peak to peak current ripple on AB out = 309mA

4B) long and grouped wire cable + external ind (330uH) + AB and DC channel on -> peak to peak current ripple on AB out = 437mA

(worst case)

5A) long and twisted wire cable + external ind (330uH) + AB channel on -> peak to peak current ripple on AB out = 118mA

5B) long and twisted wire cable + external ind (330uH) + AB and DC channel on -> peak to peak current ripple on AB out = 118mA

(best case)

6A) long and twisted wire cable ( AB & CD wires grouped)  + external ind (330uH) + AB channel on -> peak to peak current ripple on AB out = 158mA

6B) long and twisted wire cable ( AB & CD wires grouped) + external ind (330uH) + AB and DC channel on -> peak to peak current ripple on AB out = 189mA

(intermediate case)

7A) long and twisted wire cable ( AB & CD wires grouped)  + external ind (330uH) with inverted channels + AB channel on -> peak to peak current ripple on AB out = 77mA

7B) long and twisted wire cable ( AB & CD wires grouped) + external ind (330uH) with inverted channels + AB and DC channel on -> peak to peak current ripple on AB out = 193mA

(PROBLEM SWAPPED ON CD BRIDGE, like the boards on the field).

I think we can conclude thath the cause is the harness.

Let me know if you have some suggestions.

Have a nice day,

Sergio

Hi Sergio,

Thank you for such a thorough evaluation, and narrowing down the cause to the harness.

I will contact a colleague who has more experience dealing with longer cables and have him reply. It may be a couple of days.

Hi Rick,

glad to read your words.

I have prepared a table where I have summarized my test's results. I hope it will be useful.

DRV8412 cable tests.pdf

Apart from the harness effects, I think now remains only one point to clarify: why the activation of the secon bridge increases the current deformation on the first bridge. Let me know what do you think about it.

Thank you for your kind help,

Hi Sergio

Thanks of your great test summary on this topic. I have some experience on the spikes induced by the parasitic capacitance load, in which it could be more serious with long wires.

Please refer to the following test.

1.  DRV8824 (Actually it dosen't matter what the driver is) drives a big inductance solenoid.

2. Drives a mid inductance stepper

3. Drives a small inductance motor (0.5mH)

Then, just add a 2nF cap in parallel with the small inductance motor and redo test 3 with exactly the same settings. We get the following result.

The waveform we seen may caused by the quick charge and discharge current during the PWM edge.

The spike should be no bad effect to the motor running.

Hello Wilson,

thank you for your contribution.

I ask you some more details about your experience; did you see the spikes on both motor windings or I suppose you seen it only on the windings with the capacitor in parallel?

Moreover, you solved removing the capacitor? in my case I haven't a real capacitor, so I can't walk this way.

I add a schematics of the loads I have to drive:

loads.pdf

Another important aspect of my case is that the presence of the deformation increases the peak current value and the RMS current value; this means that I have a misalignment respect the clean channel behaviour. The microcontroller sets the same current level (setting a 4 channels SPI controlled DAC) in the 4 output channels, but I get two channels with increased RMS current values and two channels with a lower RMS current.

In my application is crucial to have a RMS and Peak current values as close each others as possible.

Today I perform the characterization of the 4 channels, with the different types of cable and I'll post the numeric results as soon as possible.

Thanks a lot for your kind help,

Sergio

Here I report a table with the results of six measurement, all accomplished in the same way:

- one channel (one full bridge) activated and the other three deactivated.

- short and twisted cable (in this way I reduce the cable influence at minimum).

- identical inductive load used on all channels.

I have reset the board between a measure and the next.

		test A		test B		test C		test D		test E		test F		max diff on same channel
Channel	DAC value	Imax [mA]	Irms [mA]	Imax [mA]	Irms [mA]	Imax [mA]	Irms [mA]	Imax [mA]	Irms [mA]	Imax [mA]	Irms [mA]	Imax [mA]	Irms [mA]	ΔImax [mA]	ΔIrms [mA]
	300	775	702	910	741	834	670	872	708	845	683	837	674	135	71
ch1	500	961	763	1087	895	1022	828	1052	869	1016	831	1012	826	126	132
	700	1174	951	1274	1062	1219	1004	1242	1036	1195	991	1216	1010	100	111
	300	861	690	897	731	834	676	833	669	855	694	802	640	95	91
ch2	500	1038	844	1098	901	1033	839	1013	828	1035	848	1001	815	97	86
	700	1127	1009	1273	1067	1221	1010	1199	995	1233	1035	1202	1000	146	72
	300	928	753	946	771	936	762	863	696	932	765	886	718	83	75
ch3	500	1104	905	1130	934	1121	925	1035	845	1106	919	1064	876	95	89
	700	1275	1059	1302	1090	1297	1084	1205	998	1283	1079	1252	1045	97	92
	300	818	641	932	758	943	769	852	683	885	718	877	710	125	128
ch4	500	1005	802	1126	929	1136	938	1030	836	1076	886	1061	873	131	136
	700	1213	990	1309	1095	1314	1101	1230	1019	1269	1064	1257	1049	101	111
		max diff between channels		max diff between channels		max diff between channels		max diff between channels		max diff between channels		max diff between channels
		ΔImax [mA]	ΔIrms [mA]	ΔImax [mA]	ΔIrms [mA]	ΔImax [mA]	ΔIrms [mA]	ΔImax [mA]	ΔIrms [mA]	ΔImax [mA]	ΔIrms [mA]	ΔImax [mA]	ΔIrms [mA]
	300	153	112	49	40	109	99	39	39	87	82	84	78
	500	143	142	43	39	114	110	39	41	90	88	63	61
	700	148	108	36	33	95	97	43	41	88	88	55	49