DWM1000 Power Voltage stability

Hello all,

We are designing modules which incorporate UWB DWM1000 modules for ranging estimation using TWR.
One module is powered by a battery through a LDO regulator. This module is used as a Tag.
The other module is powered from vehicle voltage (9V to 48V) and using a switching DC/DC regulator. This module is used as an Anchor.
NB: in the experiments below, the Anchor is powered directly from a Lab Power supply; we did not use the switching DC/DC as we encountered problems with the one implemented.

On both modules, we observe voltage drops when the transceiver is activated (and mostly in RX).

On the anchor side, which listens for incoming packets using Sniffing mode, we observe voltage variations. Same behavior is observed on the current (but no screenshot).
Sniffing configuration is 64µs Off and 64µs On, and is perfectly coherent with the voltage/current variations.
I cannot add the screenshot, if needed, I’ll find a way to transmit it.

On the following screenshot, we observe the voltages of the two modules.


Yellow trace is for the Tag and Green trace is for the Anchor. The sequence observed is the complete exchange for TWR using three messages plus an extra message.
Each time one of the module transceiver is turned on in RX, the voltage drops. The drop on the Tag side is small. However, on the anchor side, the voltage drop is more important.

Do you have some insights on how to build the power supply for the DWM1000 to avoid such behaviors?
Do we need extra capacitors, not all of the necessary are included in the module ?

Thank you for your help,
Best regards,
Aleks

You’re always going to get some sort of voltage drop when the current increases, you just have to do your best to minimize them.

How is your power supply currently connected? Does it have capacitors on the required output pins and then a low impedance connection to the module?

I can’t see and specific requirements listed for the DWM1000 but you should treat it like any other IC and put a bulk capacitor as close as possible to the power supply input pins. Given this is a fairly high current module which should contain the required smaller power caps close to the pins I’d say something in the 1-4.7 uF range would be best.

Hello AndyA,

Thank you for your insights!

The voltage drop on the Tag side looks acceptable. And I agree with the voltage drop due to the current increase but on our Anchor side, the drop is quite important and enters in the need of minimizing it.

The power supply is wired using lab wires and crocodile clamps. When using the embedded power supply, the recommended capacitors are used around the regulator.
I did not do the routing myself but after checking, tracks seem to be of low impedance.

We did not see any requirements either for the DWM1000 and that is why we did not put any bulk capacitor on the power supply pins.
With your suggestions, I tried adding a bulk capacitor of 10µF (that’s what I have in stock) and this effectively reduced the voltage drop by around 30%. Power supply decoupling goes in the right direction for solving this problem.
Interestingly, the capacitor used is low ESR. I do not have any idea of how much importance this has on the decoupling efficiency, and if a lower capacitor value with low ESR is going to be sufficient?

That is not a problem about Capacitors, that is a problem about regulators!

Which IC are you using as regulator?

By the Way use a lab power supply as regulator is a terrible option even for testing purposes, that is the reason for the high level of noise that you are getting, you are not gonna solve that placing capacitors you can be sure, you should use a LDO (Linear Drop Out) Regulator with a high level of PSRR (Power Supply Ripple Rejection) Look at “TPS7A4701” that one is one of the better regulators that exist in the market for RF applications.

Hi DonQuijote,

Thank you for your feedback about the regulators. I did not know about RF targeted regulators but this seems in concordance with our observations.
We looked at the one you are recommending and are implementing it. I’ll be back with the obtained results.