What is the issue?
SwitchBlox Rev C2 sometimes does not turn on. This issue occurs sporadically when….
- The input voltage rise time is slow (longer than 10ms)
- A 0.5A or greater load is applied to the 5V output at startup.
What is causing this issue?
The issue appears to be due to the XL2011 regulator IC not behaving as expected. The issue doesn’t appear to be a mistake in the design of the schematic or layout.
Figure 1 shows the power circuitry on SwitchBlox, and figure 2 shows the example schematic in the datasheet for the XL2011.
The only major difference is the addition of a voltage divider on the FB to increase the output voltage slightly to 5.2V. We initially thought this could be causing the issue, however even when removing this circuitry, we saw the same issue. We have tested the XL2011 on a breadboard using a circuit exactly as in the datasheet, and see the same issue.
Therefore we have concluded that this issue is not the result of a design flaw on our side, but rather is a fault internal to the silicon of the XL2011 IC. This may be a fault in the silicon, or just a poorly designed IC.
How did this issue slip past testing?
This issue is particularly unpredictable and difficult to find. Sometimes it never occurs and other times it occurs randomly. The issue will usually show itself when using a power supply with a slow rise time, and applying a 0.5A load immediately at startup.
Our manufacturer conducted power-on tests, and we conducted power tests using a single power supply; the board passed both tests. Unfortunately it was only when we switched to using a different power supply that we saw this issue.
Most of the time, the board works. However “Most of the time, it works” is not an acceptable quality for our hardware.
What is the fix?
Currently there is no way to fully fix this issue on SwitchBlox Rev C2. We can apply a manual fix that will solve this issue, however this manual fix also removes the 5V output, and reduces the maximum input voltage to 30V. For customers where this is acceptable, we will be offering this fix.
This manual fix involves bypassing the XL2011 and replacing the downstream 3.3V regulator with a higher input voltage variant from TI (TPS70933DBVT). This allows the board to work from 5-30V, but disables the 5V output.
Ultimately there is no way to fix this issue without revising the design to use a better regulator. We have already completed this revision, (Rev D) and it is currently in manufacture.
Which boards does this affect?
This issue only affects SwitchBlox Rev C2. It does not affect any previous versions of SwitchBlox, nor does it affect any other products.
How do I know if my board is affected?
If you have a SwitchBlox, check the silkscreen label on the front side. If the text says Rev C2, then this issue will apply to your board. If you haven’t already been contacted by us, please get in touch by emailing email@example.com.
How will BotBlox address this issue?
We are very sorry this has happened and will fully support our customers in resolving it. Thankfully there are only four customers affected by this, and only 32 boards have been shipped with this issue. We will contact these customers directly and are offering the following options.
Option 1) Immediate shipping of fixed boards + free upgrade
- We will ship replacement boards with the fix described above. These boards will work with an input voltage up to 30V, but with no 5V output. The lead time is around 3 days + shipping.
- We will ship the revised version of the board (Rev D) when it has completed manufacture and testing. The lead time is around 30 days + shipping.
- This service will be offered at no fee to the customer, and the customer does not need to ship any boards back to us.
Option 2) A full immediate refund, with no obligation to ship the boards back to us
Are there any other workarounds?
Yes, see below.
- Workaround 1: If your system has a 4-6V voltage rail, you can apply that directly to the 5V Output of SwitchBlox, and it will power on as normal. This approach bypasses the higher voltage DCDC converter and feeds the 3.3V regulator directly.
- Workaround 2: This is the manual fix we will apply, which bypasses the DCDC converter and replaces the 3.3V regulator with a version that is tolerant up to 30V.