TMS320F280039: PMBUS Data hold time after SCL fall

Hi Terry, 

Just to clarify, do you mean 300ns? That is what the datasheet says.

Best Regards,

Delaney

yes, I means 300ns, so you could confirm there should be unstable while master hold time <300ns?

Hi Terry,

It looks like that spec is the same for all of our datasheets of devices with the PMBus module. The minimum 300ns is what we would require for this module to ensure proper behavior. It may be able to operate correctly under 300ns as well, but we wouldn't be able to guarantee it. 

What value do the OEM master products use for that timing (and could you attach the datasheet for that device)? I can bring this information to the design team and see what they say / if there is some specific unstable behavior to look out for. 

Best Regards,

Delaney

Hi Woodward

Thanks your help, there are many OEM which use difference master device, customer appreciate if we could help evaluate risk for F280039 running at 200us PMBus hold time? And do we have any suggestion workaround to improve the reliability under 200us condition?

Hi Terry,

Ok, I have contacted the design team and will let you know when I get a response back from them. 

Best Regards,

Delaney

Hi Terry,

After consulting with other experts, we do have this spec value changed for a newer device - F28P55x. There is an added setting that will allow a 0ns hold time. Let me know if they would be open to switching to this device.

Also, is our C2000 device the only target on the bus / how many targets are on the bus? And which device is the transmitter when they are seeing the 200ns (or is it in both cases)? It sounds like the master device may be having a clocking issue. 

Best Regards,

Delaney

Hi Delaney

Yes, we have promoted the F28P55 with 0ns hold time to customer, they could develop this new platform but need time to complete it (usually their new project development schedule is 1 year above). The challenge is their product base on F280039 as target device already MP on the market, and have working with OEM master device which PM BUS hold time may lower to 200us.  Our F280039 is only doing target, and it maybe include 4~6 targets on the bus depend on OEM system. So far can not sure if it already happen some frame error, because system will not report if just few error, but it maybe a potential risk finally while more set system MP in market and more longer time running on the market.  So customer appreciate we could help evaluate how much risk? And do we have any suggestion workaround?

Hi Terry,

I see, yes. I am still waiting to hear back from the design team but let me ping them to see if they can help explain the risk involved.

Best Regards,

Delaney

Hi Terry,

The design team has confirmed that the risk here is that there is a possibility of invalid start or stop condition detection. With so many targets on the bus, my understanding is that there is a higher risk of noise on the bus as well, so the risk would be present.

If they have to stick to F28003x, my only suggestion would be to switch to the I2C peripheral since this also has the 0ns hold time requirement. The ideal solution would be for them to increase the hold time from the master device to 300ns since this is standard.

Best Regards,

Delaney