5V Tolerant?

Hi,

You might find XAPP646 on the Xilinx website helpful. There are a number of ways you could approach the problem you are facing. I don't know if the 100 ohm resistor approach was designed with the PCI bus in mind. Instead you may want to consider "bus switches" which seem to be the most common solution for this problem. Either way, use of external parts like this is not compliant with the spec. Some people demand full compliance.

If you have the option, I personally think the best thing to do is use Spartan-II (instead of Spartan-IIE) because you can connect it directly to a 5V PCI bus. Sometimes this is not possible due to pricing, or other features you need, but it eliminates external components and is compliant.

Eric

Followup to: By author: "Brian" In newsgroup: comp.arch.fpga

Most of the cases I've seen of 3.3 V designs connecting to 5 V PCI have been using quickswitches rather than resistors. See for example Xilinx appnote XAPP646. You can get them quite wide these days, so you don't need too many parts, even.

-hpa

If you are designing a custom card that won't be sold for general use on the PCI bus, then you might be able to get something like this to work. But the 100 ohm resistor is going to screw up your driving ability and the bus will run slowly. I expect you will need to keep your clock speed much lower than max to allow for the increased rise/fall times. But only calculations and/or simulations will tell for sure.

But why not just use the Spartan II parts which, IIRC *are* 5V tolerant???

--

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX

100

Thanks for the replies.

I'm using an evaluation board, which I'm going to connect to the PCI bus via a prototyping card. My budget doesn't really stretch to getting a new FPGA. The QuickSwitches seem to be only available in surface mount packages, which is no good for me. The card I'm making will only be used by me so using non spec parts is fine. If resistors did the job they would be great.

Someone has said that the 5V PCI bus in modern PC's actually uses 3.3V signals. If this is the case, then I may not need any level shifting. I'll measure the PCI voltages later today and post the results here.

How can I calculate whether or not the 100 ohm resistors will reduce my clock speeds? Am I right in assuming I need to do a SPICE simulation? If so, can anyone recommend some free SPICE software (Linux is preferable to Windows)?

Thanks again.

I'll

Measuring the PCI voltages won't give you what you want. There are separate pins for 5V and 3.3V and both may be present. A better indicator would be where the keys are located in the bus connectors. If the key is near the front (connector end) of the board, it is a 3.3V slot. If on the other end of the connect it is a 5V slot. Universal PCI cards have a slot in both locations so it can plug into either type slot.

--
Greg
readgc.invalid@hotmail.com.invalid
(Remove the '.invalid' twice to send Email)

Hi Brian,

Quite honestly, if I were you -- I'd sell all the stuff on eBay to generate a downpayment for a PCI development board from

which costs less than $300.

You have to ask yourself -- what is your time worth? What are the chances you might build something that doesn't work properly? And, if you are unlucky in that regard, what is the cost of failure?

The switching thresholds are similar to 5.0v TTL, and are compatible with 3.3v CMOS devices. That's why with Virtex and Spartan-II, you can interface to 5.0v PCI using a VCCO (output driver voltage) of 3.3v. When you use those parts, though, there are no clamp diodes to 3.3v in the FPGA I/O.

That is not the case with Spartan-IIE. The clamp diodes in the FPGA to 3.3v will be turned on by the pullups on the bus to 5.0v. This is not a good arrangement.

This is a lot of work with SPICE. Even when you arrived at a result, "I can only run the bus at 25 MHz" there are few machines that will let you change the bus frequency. So this work won't buy you much if you're on a budget.

Good luck, Eric

There is another possible meaning for "voltage". The levels of the signals on the data/control pins.

For a one-off hack, I'd put a scope on the data bus and see if anything goes over 3V. (I did that on one system many years ago. It was a 5V system, but the signals never went above 3V.)

Of course, you can shoot yourself in the foot if you add another card that does drive the data bus to 5V. Or didn't test things carefully enough and missed a board in your system that does it.

--
The suespammers.org mail server is located in California.  So are all my
other mailboxes.  Please do not send unsolicited bulk e-mail or unsolicited
commercial e-mail to my suespammers.org address or any of my other addresses.
These are my opinions, not necessarily my employer's.  I hate spam.

separate

end

unsolicited

addresses.

Yes, I hoping the signal levels (data, clock, reset, etc) never go above

3.3V. Even though the PCI system is 5V. I've lost the probes for my scope, so I can't check it until I get some more. The card is only a one off so this quick fix should be okay.