AD243x series replacement for AD2428

Hi matthijs 

For non-automotive applications, the available A2B transceivers are AD2426, AD2427, AD2428, and AD2437. Please consider using the above parts in your design, and let me know if you face any challenges in meeting your application requirements.

The AD2437 would be fine if it can be operated in a manner compatible with the AD2428, with data and 4-9V bus power over a single twisted pair, but I find the documentation ambiguous:

• The product page and datasheet mention "The transceiver supports a bus power feature, where the main node supplies voltage and current to the subordinate nodes over the same daisy-chained, single-pair wire or XLR/DMX cable as used for the communication link." but no reference design seems available for the single-pair use case.
• The operating conditions in the datasheet says minimum VBUS is 12V, but then in footnote 4 says in LVI mode nominal VBUS = VIN = 3.3V.
• Even when 12-24V bus power is used, it looks like low voltage (5V) bus power is still used initially during discovery.
• It appears the VBUS input is only used for monitoring, and that the thresholds for this monitoring are freely programmable (via the VMTR registers), making it unclear why it would have a 12V minimum.

Can you clarify whether 4-9V VBUS can be supported by the AD2437, either in normal mode (VIN = VBUS = 4-9V) or LVI mode (VIN = 3.3V, VBUS = 4-9V) ?

Hi matthijs 

Please find our answers below:

matthijs said:

The AD2437 would be fine if it can be operated in a manner compatible with the AD2428, with data and 4-9V bus power over a single twisted pair

AD2437 and AD2428 are not interoperable. 

• The product page and datasheet mention "The transceiver supports a bus power feature, where the main node supplies voltage and current to the subordinate nodes over the same daisy-chained, single-pair wire or XLR/DMX cable as used for the communication link." but no reference design seems available for the single-pair use case.

Ans: will get back to you.

• The operating conditions in the datasheet says minimum VBUS is 12V, but then in footnote 4 says in LVI mode nominal VBUS = VIN = 3.3V. And do not attempt for a discovery from this node onwards.

Ans: VBUS can be as low as 3.3V only for the last node, if it is a locally Powered Subnode (LPS) operating in LVI mode. The voltage specified in the datasheet is required for initiating discovery and performing fault diagnostics.

• Even when 12-24V bus power is used, it looks like low voltage (5V) bus power is still used initially during discovery.

Ans: This one is applicable for XLR/RJ45 based designs. you can refer to this page for more info:Two Step Discovery for AD2437 Platforms [Analog Devices Wiki]

• It appears the VBUS input is only used for monitoring, and that the thresholds for this monitoring are freely programmable (via the VMTR registers), making it unclear why it would have a 12V minimum.

Ans: BUS is essential for line fault diagnostics and next node discovery. The values/thresholds are not configurable as those are hard wired in the design, VMTR ADC will not affect the diagnostics/discovery though it can be used to determine if the VBUS (or other supplies) are going beyond or above the user defined threshold. Even if you don't enable VMTR ADC the discovery/diagnostics will work as intended. 

If you don't mind managing bus power yourself then you can disable/ignore the AD2437 bus power management features. ADI does this on their own RJ45 scheme where 5V is carried over the A2B pair and the 24V is on separate wires.  Gotta warn you that the AD2437 chip is not well documented and there might be some trial and error involved.  But it can and does work as I've built hardware with AD2437 and 5V over UTP.

arunk said:

• no reference design seems available for the single-pair use case.

Ans: will get back to you.

Thank you

arunk said:

BUS is essential for line fault diagnostics and next node discovery. The values/thresholds are not configurable as those are hard wired in the design,

Understood. So my initial assessment was correct, the only AD243x device that's a backwards compatible upgrade for the AD2428 is the automotive AD2433W (which is documented to use power config 0) ?

Many of the AD243x features would be very useful for us, including

• lack of power supply sequencing constraints
• SPI support
• Faster I²C
• TDM crossbar

However we've already designed a product with the AD2428 and we need interoperability. It seems a bit unfortunate that Analog isn't providing a backwards-compatible upgrade path to the AD243x series for non-automotive users.

What exactly is the obstacle to using the automotive parts in non-automotive applications?

Oh wow, I hadn't looked in detail yet at how the AD2437 TRM handles bus power and discovery for the XLR and RJ45 reference designs, but yeah you're right they're setting CONTROL.SWBYP=1 during first discovery phase.... which also means they sacrificed pretty much the entire bus diagnostics feature (everything becomes "timeout during discovery").

So you're just always using CONTROL.SWBYP=1 and control your bus power via GPIO?

And yeah I'm also not super happy about the AD2428 docs, e.g. I really wish the exact trip conditions for the various bus faults were better documented, including how they're affected by SWCTL.MODE and SWCTL.DIAGMODE, especially since in our application we can't avoid ground interconnection between devices.

The AD2437's bus power scheme is still used for the 24V, but it's not turned on until it's been verified that the downstream chip is another AD2437 and its EEPROM has been programmed with a  bit that says "safe to give me 24V".  Once that's been completed then the rest of discovery completes and power fault monitoring is active.

This prevents blowing up some non A2B thing (or an older A2B CFG0 thing) by applying 24V to it.

The normal UTP things don't do that check as in an automotive use case everything is fixed.

But there's tests I've done that show disagreement with the TRM, as well as when you dive really deep into the TRM you'll see there are many bits that control things that are not explained.  I've worked though those by tracing the software and an I2C analyzer.

The ground issue you point out is real, that's why the XLR and RJ45 hardware doesn't have sensing on the GND (I like to call it Return as GND implies some mystical thing that doesn't exist).

Another way I've solved problems is to galvanically isolate A2B from the rest of the system.  Not always practical but it does work.

One other minor point. My memory is the AD2433 I2S/audio matrix is a bit different from AD2437. I didn't go back to their respective TRMs to remind myself of exactly what.  Because I do some automotive work too I've used AD2433.  I can't pretend to understand ADI's rules for how/why they sell parts.

Not to turn this into an ad and break some forum rule(s), but my company (Clockworks Signal Processing) does A2B development/consulting.  Feel free to reach out directly to me.

Brewster said:

The AD2437's bus power scheme is still used for the 24V, but it's not turned on until

I know, that's why I said "first discovery phase". It still means the bus diagnostics are unavailable during discovery, which is when you need them the most.

Brewster said:

The ground issue you point out is real, that's why the XLR and RJ45 hardware doesn't have sensing on the GND

I find it interesting though that they pull the sense lines to ground. From my testing it seemed the AD2428 needs to see current flowing into VSSN to avoid getting a power fault. What seemed to work consistently is making sure a node's A-port always presents the standard dummy load across the bus, and hook up the node's ground via a diode:

That way if there's an alternate ground path the upstream node still sees the dummy load, which seems enough to keep it happy even if most of the bus current uses an alternate path that bypasses VSSN.

Brewster said:

I can't pretend to understand ADI's rules for how/why they sell parts.

Yeah I'm a bit puzzled why they market a select few A²B parts for general purpose use, and have public documentation for those, but continue to act all hush-hush about the rest of the family instead of just making the whole portfolio available (even if it might mean you pay a slight premium for automotive-qualified parts). Presumably it makes some kind of business sense to them, but it's a bit annoying to see a part that seems to be what I need but get told "no, you can't use it because the product you're making isn't a car".

Brewster said:

Clockworks Signal Processing

Ah, yes I recognize that name, I've read some of your technotes.

There's a subtly on ADI's AD2437 RJ45 and XLR implementations.  Whatever you use for the 5V on the A2B pair should have a "fault" output.  ADI uses 50 mA but I use 100 mA for some specific reasons.  That fault line ties to GPIO7 for using ADI's default driver code.

That will tell you you have a short.  The other wiring faults would cause the downstream node to fail to reply to the discovery probe (and don't get me started on how I think ADI made a big oops on that...) and that's all you really need to know (broken or not broken) as you would not apply 24V if you come up broken.

The return current sense thing is different on AD2437.  As I mentioned ADI does not use it for RJ45 and XLR.  It works as expected on UTP with the AD2437.  Though either way it can be disabled.  AD2428 I'm not sure I've tried to cheat.

Another heads up on AD2437 is look at the Rsense values.  They don't make sense (no pun intended).

Hi matthijs 

Kindly use AD2428 for UTP design and AD2437 for other wiring applications. Although I agree that some features are not available in AD2428 compared to AD2437,

I can see some conversation regarding the design has been ongoing in this forum chats. Please note that the parameters should stick to the datasheet values to ensure robust performance. If you have specific questions regarding any portion of the design, please post them here, and we will take them up.