USB interfacing, Using a DLP245PB, I get very bad ringing on the data lines

more likely your construction is "far too simple" to prevent this type of ringing, which is most likely an LRC circuit phenomenon (3 passives - simple enough?).

inductance L = a loop of wire (L is proportional to area of loop), R = R of wire, C = input Z of gate etc. Voila, drive it with a "square" wave and watch it ring.

build your circuit a bit better, and watch the hairy mess go away.

What do you mean by "ringing". Do you mean oscillation?

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

What are you using to measure the 'ringing'. While an oscilloscope is implied, I am curious about your probe. A 10x or a 1x probe can cause signals to appear to ring by the application of the probe. This is similar to the proverbial problem of where applying a probe makes the circuit work that otherwise doesn't. As someone else pointed out, the ringing is the result of an RLC circuit. Depending on way you are measuing you can induce a hell of a lot of L (in the ground) and some C from the scope probe making a very high Q circuit.

You really need to look at it with either an active probe ($$$) or if your scope has a 50Ohm termination capability, you can use a piece of

50Ohm coax and attach a 1k resistor to it to make a 20x (attenuation) probe that is almost purely resistive.

If I understand correctly, you are seeing the ringing on on the clock line, which I am guessing is casuing multiple clocks on your epot. Could you elaborate further as to the difficulties you are experiencing?

If my assumption is correct and that it is multiple clocks due to signalling on the clock line, my next set of questions would be in regards to the circuit implementation.

How is the clock circuit constructed, ie is it on a PCB, it is a bread board with wires, wirewrap, etc?

Also, very important how long is the clock line run and what is driving it? If it is more than a couple of inches long, you will likely need to terminate it as a transmission line.

translation: go to linear technologies website and read AN47

Cheers Terry

Hi everyone, I solved my problem, it has more to do with inexperience than bad design

The advert on the FTDI website advertises the DLP245PB

as having 18 I/O lines, and an 8-Bit data bus. I mis-interpreted this to mean that there was a total of 26 different data lines that could be used, 8 of which was accessible by this data bus. I assumed that this meant that the 8-bit data bus would be faster, or provide better throughput, or some other strange idea (I am not an electrical engineer, don't judge me harshly please)

It turns out that the 8-bit data bus, although it occupies pin-space on the DLP245PB DIP board (pins 21 through 28 of the board), and is connectted to PORT D on the PIC16f877, it is not supposed to be used by users. It is the direct connection between the FTDI245 USB interface chip and the PIC16f877 that is on the board.

I am not sure why they designed this board like this, but I guess that it means that people who want to program the on-board PIC (change the on PIC software, whatever) and would like to be able to access pins in order to ascertain wether their PICFTDI communication protocols are functioning correctly would be able to have this option in order to be able to check and make sure they had programmed everything correctly. The other possibility is that it is a secretive bit of underhandedness meant to lure in the unwary/naive who thought that the more pins available for data access the better, but that's just conspiracy Mike thinking out loud.

In short, the ringing I was experiencing on PORT D (Pins 21-28) was in fact the chip-to-chip communication between the PIC & FTDI usb interface, and not signs of a more serious circuit flaw. This is why my clock kept skipping ahead 5 beats for every tick I was making, and why filter capacitor solutions made the device stop working.

The solution!!! : Move the pin functionality to PORT B (Pins

29-31), modify my software accordingly, and BOOM, no more problems.

One other note, is there a way to post files in here (I am writing a help document on the use of the DLP245PB as part of an upcoming echnical writing class I will be taking during the summer, and I would like some feedback on it, it is a beginners guide to the use, and programming of the DLP245PB for novices interested in PC interfacing using the serial port.

ax snipped-for-privacy@yahoo.com wrote:

The data bus design of the DLP245PB is also why PORT D can be used for blinking LEDS, and for other types of Slow/Imprecise applications, but should never really be used for interfacing to IC's

IT also explains why hooking up stuff to PORT D on the DLP245PB will also start to degrade the operating performance of the DLP245PB, resulting in more hang ups and slow downs. Leaving PORT D alone and unconnected (don't touch pins 21-28) wind up solving the problem and helping performance.

By the way the help file I am writing is currently in word format. Is it possible for me to post some of that here?

And thank you all for your help. I am the original Ax_deimos guy who started this thread and I thank you all for your rapid responses.

No - "here" is USENET - all newsgroups are by default text-only unless they have "binaries" (or some misspelt version) in their title. e.g alt.binaries.schematics.electronic A raw .doc file is binary. You may either post it in a binary newsgroup or - probably easier for everyone - put it on a website somewhere and include the url in your post.

Geo

I agree that their site info is misleading:- "18 digital I/O lines ( 6 can be configured as A/D inputs ) plus the 8-bit data bus available for interfacing to user electronics"

To me that would indicate that the databus /is/ available. Looking at the circuit, I do not really see how it could be used, since it will be carrying the commands to the PIC from the pc.

Geo

In addition, many people are very shy about opening word files due to the nasties that are sometimes hidden in them. I would suggest pdf (word to pdf converters are available for free on the net)

David

Geo wrote:

have

-

If it's plain ascii. You can simple mark, copy and then paste into your posting. To rid of all m$-junk.

Url's are nice. But not for anyone looking at the post more than a year later on. As websites tend to dissapear.

I got this info from Don Powrie, one of the Tech staff at DLPDESIGN

At this point I think I understand why this was set up in this way. It still leaves me with 18 data lines to use with other projects

on the

"

Ok - so you could add an 8 bit latch o/p latch needing a strobe line and an 8 bit input buffer with tri-state control line, regaining a net 6 i/o bits.

Geo