USB UART board problem

[ion]

Hi

A few months ago I made a USB to UART board using the FT230XS chip.  The circuit is based on some open source designs and the FTDI datasheet.
It didn't work.
I get power - not surprising as it's connected directly to the USB 5V and GND.  I could also detect the signal from the micro making it to the IC, but nothing on the USB side.

I wouldn't be surprised if the problem is the board itself as this was one of my first attempts at etching my own pcb.  In fact I tried to reflow the chip a couple of times and some of the traces have disintegrated.

I recently managed to etch a board that works, so I thought I'd have another go at this one.  However, I am not 100% confident that I haven't somehow messed up the circuit.  So if anyone can see anything wrong in the attached schematic, please let me know.

C1:              2.2uF
C2 & C3:      0.1uF
R1 & R2:      27R

[David_AVD]

First, check the reset pin info in the data sheet.  I'm fairly sure it can be left floating or tied to VccIO.  You have tied it to +5 but your VccIO is 3.3V so not sure if that will have done any damage.

Also post your layout.  The USB side is not super critical (I've never had an issue with FT232BL/RL) but you still need to observe basic rules for routing and decoupling.

[suicidaleggroll]

You should make sure you're not suffering from serial dyslexia as well.  Tx and Rx always have a perspective.  WHO is Txing and WHO is Rxing.

The FT230XS chip is labeled from its own perspective.  Tx on the data sheet means the FT230XS chip is Txing on that pin, and Rx means the FT230XS chip is Rxing on that pin.  You need to tie the FT230XS Tx to your micro's Rx and vice versa.  I always find that arrows are a much clearer way of communicating the direction of comms versus the ambiguous "Tx" and "Rx" labels that everybody uses.

[Sebastian]

Stupid question, but do you have the right driver installed on your PC?
Your PCB layout could also be helpful to troubleshoot the problem, even though it does only full speed USB so it's not critical.

[ion]

First, check the reset pin info in the data sheet.  I'm fairly sure it can be left floating or tied to VccIO.  You have tied it to +5 but your VccIO is 3.3V so not sure if that will have done any damage.

Also post your layout.  The USB side is not super critical (I've never had an issue with FT232BL/RL) but you still need to observe basic rules for routing and decoupling.

Interesting - unless I missed something the data sheet says tied to Vcc or floating, but then all the diagrams show it tied to VccIO.  I tried cutting the track to leave it floating but that didn't help.  Hopefully I didn't damage the IC - a quick look on Google suggests it's 5v tolerant on all pins.

You should make sure you're not suffering from serial dyslexia as well.  Tx and Rx always have a perspective.  WHO is Txing and WHO is Rxing.

The FT230XS chip is labeled from its own perspective.  Tx on the data sheet means the FT230XS chip is Txing on that pin, and Rx means the FT230XS chip is Rxing on that pin.  You need to tie the FT230XS Tx to your micro's Rx and vice versa.  I always find that arrows are a much clearer way of communicating the direction of comms versus the ambiguous "Tx" and "Rx" labels that everybody uses.

This could be at least part of the problem.  The board is layed out from the IC perspective, but I could have easily hooked it up from the micro's perspective.  Pretty sure I tried swapping the data lines while troubleshooting at some point so it wouldn't be the only problem.
I like the idea of adding arrows though.

Stupid question, but do you have the right driver installed on your PC?
Your PCB layout could also be helpful to troubleshoot the problem, even though it does only full speed USB so it's not critical.

I tried that, didn't help.  I'm sure if it was a driver issue the PC would still detect something on the USB port right?  I wasn't even getting an unknown device!


I'm attaching the board layout.  I colour coded the pads for the caps as they're not a standard footprint - I'm using through hole caps soldered to smd pads.

[Sebastian]

The 47pF edge rate limiting caps are missing. I don't have a lot of experience with stuff like that yet, but I think that could lead to signal integrity issues. The power rails crossing your USB data lines is not the best thing to do and a ground plane underneath the data lines would also be good but probably not essential. There should also be a ferrite bead on vcc after the connector but i think that won't affect the function, just EMI. Reset should be tied to VCCIO which is 3.3V, not VCC which is 5V. You could have damaged the device by doing that, so if you have a spare one you should swap it after you changed that, just to be sure. C3 could also be close to the chip especially if it has long leads, a SMD cap would be better.

I know that is a lot of stuff but I don't think it is very critical and I think your layout could still work but you have to change the reset pin and maybe replace the IC.

[ion]

IIRC the open source design I used as a reference didn't have the 47pF caps or ferrite bead, so I guess it should still work without them.  Might as well add the pads for the caps though seeing as I have to modify the layout.
Yes, the layout is far from ideal, I really wanted to avoid using jumpers.  And I didn't really know what I was doing at the time.

Seems everything you mentioned other than the Reset connection has to do with signal integrity.  I wasn't getting any signal out of the chip.  I've got a bad feeling I'm going to have to get a new IC.

I'll update the layout and hopefully should have the time to etch the new board in the next week or two.  Will report back with the results.

[suicidaleggroll]

If you're going to be waiting a couple of weeks anyway, why not order the PCB online?  You could probably get that board manufactured for less than $2 each, including solder mask, silkscreening, and you could make it double sided with all of the necessary vias, copper pour, etc. to optimize your layout.

[David_AVD]

IIRC the open source design I used as a reference didn't have the 47pF caps or ferrite bead, so I guess it should still work without them.

Did you read the FTDI data sheet for the chip?  All the ones I've seen have many examples and go into quite a bit of detail for different scenarios.

[ion]

If you're going to be waiting a couple of weeks anyway, why not order the PCB online?  You could probably get that board manufactured for less than $2 each, including solder mask, silkscreening, and you could make it double sided with all of the necessary vias, copper pour, etc. to optimize your layout.

Because this is mostly a learning exercise.  This board is something that's nice to have rather than something I urgently need, and I've wanted to try etching my own boards for a while too.
I'd rather refine the process on something non critical as I may actually need to knock out a few quick prototypes in the near future.

[ion]

IIRC the open source design I used as a reference didn't have the 47pF caps or ferrite bead, so I guess it should still work without them.

Did you read the FTDI data sheet for the chip?  All the ones I've seen have many examples and go into quite a bit of detail for different scenarios.

I did when I was initially laying out the board.  The ferrite is for EMI so I guess I can leave it out.  I can't find anything in the datasheet refering to how critical the 47pf caps are, but as I said I had seen a design without them so at the time I figured they weren't that important.
No reason I can't add them though.

[Monkeh]

Put the caps and the bead in.

While you're at it, have a seach for USB layout guidelines. PTH resistors don't want to be there.

[David_AVD]

Just because you see a design on the web that's missing parts from the app note, don't take that as meaning the design is good or ever worked.  The web is full of half arsed projects that never went beyond the drawing stage.

[Sebastian]

I am not sure if the USB host controller has it's own edge rate limiting if it is in full speed mode, but if not the 47pF caps would be essential if the capacitance of the traces and cable is not high enough. With the rise times of high speed USB which is 480MBit/s I think, the harmonics are in the few GHz range which could, depending on the length of them, make your traces act like transmission lines and you are going to get reflections. If the traces are short you could probably still get away with it though.

[ion]

Put the caps and the bead in.

While you're at it, have a seach for USB layout guidelines. PTH resistors don't want to be there.

Seems like I'll have to make quite a few changes.  Didn't even think to look up layout guidelines, thanks for that!
I can see now that it's a good idea to put the ferrite in, but would I be right in saying that it would still sort of work if I put a jumper across the pads (for testing purposes)?  It's just that sourcing one will have to wait till my next online order and I'd like to confirm if I should add a new IC to the list.

I am not sure if the USB host controller has it's own edge rate limiting if it is in full speed mode, but if not the 47pF caps would be essential if the capacitance of the traces and cable is not high enough. With the rise times of high speed USB which is 480MBit/s I think, the harmonics are in the few GHz range which could, depending on the length of them, make your traces act like transmission lines and you are going to get reflections. If the traces are short you could probably still get away with it though.

Thanks for the explanation, I guess there's no reason not to add them.

Just because you see a design on the web that's missing parts from the app note, don't take that as meaning the design is good or ever worked.  The web is full of half arsed projects that never went beyond the drawing stage.

Good point!  But the schematic I saw was from a board that is for sale, so I figured it had to work.  Pretty sure I saw it in a Mailbag video a few months back.

[Monkeh]

Oh, it'll work without a ferrite, but generally if they're specified you should use them.. They're cheap enough from Farnell, stock up on some.

I'd definitely get at least one replacement IC. And re-read the datasheet. These chips are pretty easy to implement.

[ion]

If anyone was wondering whether or not 5V on the reset pin damaged the chip, it seems I won't be able to answer that.
I somehow managed to damage one of the IC pins while desoldering it (and yes, it had to be one of the USB data pins!)   

Well, I guess that means the project will have to be shelved until I get a new IC.