PL2303 serial adapters

Should I assume you know about the counterfeit Prolific adapters? It's been awhile since I've looked at the issue, but there are lots of them out there. I believe the Prolific drivers won't work with them though, so it sounds like this is not your problem. Dig around a little and see what the current status is.

At one time Prolfic released a driver that would brick counterfeit adapters. There was a lot of blowback and they pulled those drivers and released some that simply wouldn't work with the counterfeit units.

Are you confusing Prolific with FTDI?

That's possible. Was it FTDI who released the damaging driver?

Even so, I know the Prolific devices are counterfeited widely which can possibly be the problem here.

I use the CH340 from wch.cn. It is inexpensive, it works and I'm pretty sure there are no knock offs.

That would be hilarious...

I knew FTDI had pulled that sort of trick (twice?), but didn't know Prolific had too ...

I've had fun with PL2303 (Stardot brand form CPC) adapters as well. Plugged into a HP Microserver and running CentOS. The one for the UPS would drop off, but immediatly come back. Trouble is NUT doesn't notice and tries to talk to the old handle that now doesn't end anywhere useful, its complaints flood the log...

Interval between "disappearences" was weeks rather than hours. I have two others on that machine one attached to a VFD display and another to a automatic weather station. The weather station one I've never noticed disappear, the VFD one does but less often than the UPS. I've swapped the UPS and Wx ones and the "fault" stays with the UPS.

I have a feeling these crashes/disappearances are somehow related to the serial device attached, maybe it sinks a wee bit too much current just occasionally for some reason. I've since replaced the one on the UPS with a FTDI UC232R-10 (FT232RL based) but that's not been in service long enough to say if the problem has gone away.

Blimey, is that just the USB or the serial side as well? Thinking their might be enough power "leaking" from the serial devices outputs to keep the thing alive enough so it doesn't forget its bad state.

Just a thought. Where are these in relation to mains supplies?

At work we found some chip based interface converters were very sensitive to EMI - especially those in plastic cases.

I've only ever seen issues with Prolific devices, genuine and clones. Clones being by far the worst.

From my experience, genuine FTDI devices work and stay working (Windows/Linux based). Based on issues and nonsense over the years I now buy only genuine FTDI based cables from reputable dealers. Costs a bit more for the hardware, saves much, much more in my time.

YMMV and cue FTDI haters whose "will never specify FTDI due to their antics" but were never in a position to influence purchasing anyway!

The USB side was what locked up, I didn't have the power lead connected on the serial side but think I never did that particular test with TX and RX disconnected. I had to disconnect all serial-side wires to plug the adapter into my Mac (different room).

A recent software update seems to have improved the behavior of the PL2303s considerably; the troublesome units have held a connection for over 24 hours, which is a record. I thought the code was stable, but maybe not in FreeBSD-ARM. It does not look like uplcom changed, but there has been considerable work on the USB side.

Thanks to everybody for taking the time to read and repl!

bob prohaska

That was how the whole issue has arisen. Designers specified an FTDI device, purchasers bought clone devices to that specification, FTDI decided to brick the devices. Designers who did nothing wrong saw their devices being bricked and decided not to specify FTDI anymore.

Perhaps it's because I've only worked for ISO9001 type companies for the last 25 years but if designers specified FTDI and purchasing changed the part without approval and it failed then the purchasing people would be looking for new jobs. Likewise if the supplier supplied counterfeit devices through ignorance or a lack of care then the supplier would be picking up the bill to rework the widgets. Under no circumstance would the designers accept purchasing or suppliers pulling a trick like this. If this happens where you work, find a competent employer!

I have a genuine PL2303 device used where I work and a customer wants us to check our software works on W10 which we don't officially support. They paid us so we're checking. Everything works apart from the PL2303. The genuine Prolific device is no longer supported by the genuine Prolific driver due to so many PL2303 chips being counterfeit. Install a driver that works and the W10 auto update will update the driver to the correct latest version that doesn't work. You can't win either way. Don't say don't use W10... when someone is paying you a 7 figure amount a year for software, IP and support services and they say W10 then you do W10 or turn down all that lovely money!

The winners are the counterfeiters.

I don't work at a company that manufactures electronic devices. But what I have understood from this story is that the whole situation was not as clear as you describe there, and some of the parties involved have bought/sold/used clone devices without knowing.

That sounds like a screwup by Prolific. I don't have those issues using USB serial devices because I use them only on Linux, where the drivers don't have to be installed separately but are part of a kernel distribution. Sometimes they are supplied by the manufacturer, sometimes created by independent authors, but the addition of extra code to detect and brick clone devices has to go past some eyes that are not on the manufacturer side.

I wouldn't say that. The losers are the users.

Just past 24 hours uptime the first pl2303 dropped its connection, the second came down at around 48 hours. Both devices were stuck tight: Unplugging and replugging produced no recognition console message and no device file.

When the pl2303s were connected to a Mac running Prolific's driver, they were recognized without issue and worked perfectly when plugged back into their original FreeBSD on Raspberry Pi 2 hosts.

For lack of a better idea, I've exchanged the two failure-prone pl2303s with the two which have worked flawlessy. If the failures follow the pl2303s, then it's tempting to say they're defective in some subtle way. If the failures follow the RPI2s then the problems lie elsewhere.

At this writing all four pl2303s have been up for about four hours.

Thanks for reading, and any thoughts.

Is there a systematic writeup of the USB-to-serial device ecosystem? This thread is the clearest account I've seen but being an outsider there's surely more to it.

bob prohaska

I assume there are newer versions of Prolific devices that *are* supported? So in theory at least, you could upgrade to the newer Prolific devices which would be supported under Win10?

Can't you block that behaviour with file permissions or a Group Policy? (Not a Windows person here - as may be obvious from my question.)

Why should anyone have to buy new *external* equipment simply because a computer OS has (supposedly) upgraded?

I assume you have missed the discussion on counterfeits? What happens when counterfeit currency floods the market? They scrap the old bills and issue new ones with more protection. Why should an inexpensive serial port dongle be any different?

Not a valid comparison.

Why?