We are currently using the Ai-Thinker BU01 and DWM1000 in our anchor design, integrated with an ESP32. The system is powered via PoE, with an additional USB Type-C input option.
This is the third revision of our hardware. Across revisions, we have significantly improved the PCB layout. The current design uses a 4-layer PCB (65 mm × 65 mm), with RF sections (Wi-Fi and UWB) properly isolated. The bottom layer is a solid ground plane, and we have included selective solder mask openings to aid heat dissipation. Additionally, there is a dedicated internal ground plane to help distribute heat more effectively. The enclosure is plastic and not IP-rated.
From a firmware and functional standpoint, the system is operating correctly. However, we are observing a temperature rise of approximately 20–25°C in the module during continuous receive mode.
We measured the current consumption in this mode to be around 178 mA, which matches the datasheet-listed specification. Upon further inspection, we noticed that both the BU01 and DWM1000 modules do not provide a ground pad or exposed thermal pad on the bottom side for efficient heat transfer to the PCB.
Given these observations, we would like to understand:
Whether this level of temperature rise is expected under continuous receive operation, and
If there are any recommended design practices or mitigations we may be missing to improve thermal performance.
You can’t heat sink the module in a significantly better way. You can’t reduce the power consumption of the module significantly. Which means there isn’t a lot you can do about the module temperature rise.
You can potentially keep the temperature around the module lower by improving power supply efficiency. This wouldn’t change the temperature difference between the module and it’s surroundings but would reduce the temperature of the surroundings and so also reduce the temperature of the module.
The other option is to not use a DWM1000 and instead use a DW1000 chip directly. That gives further options for improving power supply efficiency and allows you to couple directly from the chip ground to your main ground plane improving the head dissipation. This does however involve some risk since an incorrect board layout could significantly degrade the UWB performance.
Hi, AndyA,
Thank you for your valuable time and evaluation regarding the heating issue.
We will work on improving the power supply design and evaluate the resulting temperature rise. Currently, the system efficiency is around 85%, and we believe it can be further improved to approximately 90–95%.
Once again, thank you for your support and valuable feedback.
Continuous receive mode is the highest power usage, and thus will be the highest temperature rise.
This has various effects. It will cause drift of the 38.4 MHz crystal since it changes frequency with temperature by some amount. If the system is in a high temperature gradient (either from only recently being put in receive, or from an outside temperature change), this can affect accuracy as the 38.4 MHz clock drifts. This is one reason why TWR type interactions should be quick if possible.
Even with a TCXO 38.4 MHz clock, we notice drift effects. For our anchor array, we turn on the UWB system into receive for about 10 seconds before we try to synchronize anchors to each other. This delay is to stabilize the temperatures of the parts, TCXO and UWB chip, so the time is more stable. Otherwise, we notice more slew in the time modeling in those first few seconds.
The change in temperature also have a small effect on antenna delay. This is relatively minor but measurable none the less.
A great way to reduce the temperature rise is to use one of the newer chips, DW3000 series or QM33000 series. They draw about 1/3rd the power, thus much less temperature rise.
If you make your own board, getting solid ground plane connections to the chip GND pins helps conduct heat way and thus lowers the temperature rise.
Mike Ciholas, President, Ciholas, Inc
3700 Bell Road, Newburgh, IN 47630 USA mikec@ciholas.com
Hi, Mciholas,
One of our primary concerns with the increased operating temperature was crystal drift and its impact on system performance. That is why we need to reduce temperature
To improve thermal performance and reduce heating, we evaluated a chip-down design approach instead of using a pre-certified module, as suggested by AndyA. However, implementing a chip-down solution would require a new FCC certification process, so we decided not to proceed with that approach.
We also evaluated the DW3000, which supports only UWB channels 5 and 9, whereas the DWM1000 supports UWB channels 1, 2, 3, 4, 5, and 7.
You seem to have been misinformed at some point. The DWM1000 is not a pre-certified module. If you use it you still need to do all the certifications that are required for the DW1000.
The DMW1001 is pre-certified but only when running the certified firmware. As soon as you change the firmware you invalidate the certification and are back to requiring a full set of tests.
Lower temperature is not as important as steady temperature. If your module is maintaining a constant but elevated temperature, within the operating limits, it will perform adequately for UWB measurement.
Lower temperature is always desired generally since that increases device lifetime and reduces thermal noise in the RF circuits, but these are very minor effects compared to getting good UWB measurements with a steady time base.
We’ve done 50+ chip down designs using Decawave/Qorvo chips, DW1000 to QM33000 series, and starting to design with the QM35000 series now. We have done dozens of regulatory approvals in many jurisdictions. If you want a contractor to help with a chip down design, we can do that for you. We can operate at any level from consultation with your engineers to doing the design entirely for you.
Does the Ai-Thinker BU01 module have any certifications? I don’t see any listed for it, so you might not be getting that with this module. Note that any certifications will be locked to the settings they tested under since that controls the RF performance, particularly transmit power. You might want to be sure your expectations for certification are met by this module. You want the module vendor to give you the FCC ID, and then you need to look it up to make sure they have valid test reports, if you are hoping to sell into the USA market and those that recognize FCC. This will also tell you under which rule they certified, such as 47 CFR 15.250, 15.517, 15.519, etc. Each rule has different restrictions on use.
The low bands (channels 1, 2, 3) are seemingly being abandoned by the UWB chip players. The only chip that works there is the DW1000. At one time, the EOL (end of life) date for DW1000 was sometime in 2028, which is getting close, but that may have been extended. All the other UWB chips seem to be focused on channel 5 and higher.
The low channels aren’t allowed in some jurisdictions and have restrictions in others. They do have long range, but the low bands are getting filled up with other services, too, and have more problems with interference. An example was a site using channel 3, 4.5 GHz, and it was right under an approach path for an airport. The system kept going out when a plane flew overhead. Turns out the radar altimeters on planes uses 4.2 GHz and emits RF energy in >1 KW bursts, and that was enough to clobber the UWB system. Changing to channel 5 solved that issue. UWB signals are very weak, literally limited to an acceptable noise floor for other services.
Mike Ciholas, President, Ciholas, Inc
3700 Bell Road, Newburgh, IN 47630 USA mikec@ciholas.com
This is great insight. This will definitely help us with the conclusion or moving forward to chip down approch. I will check with our client and provide your services information to them.
Once again, thank you.