AR8035 unobtainium vs TI's DP83869 Gigabit Phy Topic is solved

If you have a simple question and just want an answer.
LordeCrimson
Junior Member
Posts: 4
Joined: Thu Nov 12, 2020 6:02 am

AR8035 unobtainium vs TI's DP83869 Gigabit Phy

Postby LordeCrimson » Thu Nov 12, 2020 7:22 am

Gigabit Phy AR8035 driver task, vs a replacement driver for the DP83869HM.

Has this driver writing already been done / potential rabbit holes?
Before I recreate the wheel.... I would like to know if anyone has been down this road with success.

In short, I am following the XE-216 / xCORE-200 eXplorerKIT's schematic on a custom PCB.

The micro-controller is the XE216-512-TQ128 connected to the Gigabit Phy DP83869HM Texas Instruments instead of the Gigabit Phy AR8035 Qualcomm unobtainium, following the Texas Instruments SNLA333 app note as a guide.

Thank You
View Solution
User avatar
mon2
XCore Legend
Posts: 1849
Joined: Thu Jun 10, 2010 11:43 am

Postby mon2 » Thu Nov 12, 2020 12:29 pm

Huh?

How many millions of units were you planning to build?

https://www.arrow.com/en/products/ar803 ... .com&csi=0

Can forward more links of other stocking agents that we have used for many components.

And

viewtopic.php?t=7959

Which should be simple to source through Microchip.

However as noted on multiple postings, there is currently plenty of available stock of the original PHY around the world.
User avatar
akp
Respected Member
Posts: 466
Joined: Thu Nov 26, 2015 11:47 pm

Postby akp » Thu Nov 12, 2020 10:31 pm

Yes there are plenty. Changing to a different PHY is easy, at least for 100Mbps. It should be easy for Gbps too. If you use the ingress and egress latency you'll have to measure it again for the new phy, too.
LordeCrimson
Junior Member
Posts: 4
Joined: Thu Nov 12, 2020 6:02 am

Postby LordeCrimson » Sat Nov 14, 2020 11:35 am

mon2 wrote:
Thu Nov 12, 2020 12:29 pm
Huh?

How many millions of units were you planning to build?

https://www.arrow.com/en/products/ar803 ... .com&csi=0

Can forward more links of other stocking agents that we have used for many components.

And

viewtopic.php?t=7959

Which should be simple to source through Microchip.

However as noted on multiple postings, there is currently plenty of available stock of the original PHY around the world.
I do not know how many devices will be built.

note taken, however the MAJOR concerns here are:

1). Counterfeit components: Risky business, risky vendors. Popularity : High : Fake Threat In the Open Market = bad juju.
2). EOL with respect to five years out.
3). 100% Success rate with the DP83859 component, even during extreme EQT tests.

Sorry you did not understand my question.

Do have a great day Sir.
LordeCrimson
Junior Member
Posts: 4
Joined: Thu Nov 12, 2020 6:02 am

Postby LordeCrimson » Sat Nov 14, 2020 11:39 am

Thank You. It looked sort of easy... was avoiding recreation of the wheel.. pit falls, will be testing this weekend.
User avatar
mon2
XCore Legend
Posts: 1849
Joined: Thu Jun 10, 2010 11:43 am

Postby mon2 » Sun Nov 15, 2020 12:36 am

Hi. My preaching hat is on :)

Fully understand your position. We have millions of our widgets out in the field and have to play similar switch-to-a-different-vendor games. It is all fine when starting on a fresh design that is your own but in your case, you are working with a proven IP that is solid with a known track record. We are not audio developers but can appreciate that this ethernet PHY is proven to work correctly in all related cases and you are working with time sensitive protocols.

As there is plenty of stock available through standard distribution, my 2-bits are to start with the same PHY, same proven IP & reference layout guidelines -> start to ship the widgets while you perfect to switch to a different PHY / vendor. One of the above links that was shared offers a Micrel PHY that is confirmed to be working, a courtesy post from the developer.

I believe that you will invest weeks or more to properly validate the use of a different PHY. That is not considering the possible tools you will have to consider to validate if the PHY sub works correctly or not. Then, what if something was missed in the testing? Yet, this PHY has been confirmed to work by XMOS and very high volumes of product shipments.

Arrow is 100% legit. We have used WIN-SOURCE (Shenzhen) for a number of proto builds and they are a good supplier that is ISO certified. Not all vendors in Asia are evil. Some / most are very hard workers. Openly, the scammers are so easy to spot that it is a joke. I recall one who claims to "have more inventory than Digikey & Mouser" yet their phone number is a single line on SKYPE and is listed to be in Europe. Zero hits on their company name on Google = red flag. Yet, their bank account is on Mongok, HK. They never could share any pix of the component nor timestamped photos we had requested multiple times. However, wanted full payment up front. Sure thing buddy.

You can also search out official distributors for Qualcomm in Asia for the same component. We have done this with success on assorted parts and have found that usually the pricing is better overseas for the same device. Often these parts are from halted projects / overruns. It is possible to source 100% new and original components offshore.

For an immediate product release, rev 1 should be with the XMOS certified PHY, the rest (Rev 2) can follow when perfected.

Just my 2 bits... :)
LordeCrimson
Junior Member
Posts: 4
Joined: Thu Nov 12, 2020 6:02 am

Postby LordeCrimson » Tue Dec 01, 2020 7:44 am

Update: Spun a PCB with the XMOS 216 and 1GBIT capable DP83869 Texas Instruments PHY, coded with success. Clock skew had to be set to zero in RGMII_Const.h, along with MDIO messages in 0x0. What concerns me is the fact that the XMOS chip was burnt during MAC assignment, according to Console XBURN. I did not see any smoke, and it operates OK as wireshark is showing the MAC address I had used in XBURN.

Who is online

Users browsing this forum: No registered users and 4 guests