Programmable Devices
CPLDs, FPGAs, SoC FPGAs, Configuration, and Transceivers
20645 Discussions

Cyclone III with 5v inputs

Altera_Forum
Honored Contributor II
3,249 Views

I'm trying to connect approximately (16) 5v TTL inputs into a Cyclone III eval board, the terasic de0 [1]. These are all input towards the FPGA only. The eval board has (2) 40-pin IDC headers for the free I/O pins. 

 

I've read something about a first generation Cyclone that using the internal PCI clamps, and an external resistor to bring 5v into the board. I don't believe this will work for the III, but needed advice before I proceed to spend $$$ and time on an alternative solution. 

 

I'm also concerned about AN 447: Interfacing Cyclone III and Cyclone IV Devices with 3.3/3.0/2.5-V LVTTL/LVCMOS I/O Systems. [2] I guess I should worry about the overshoot. 

 

Of course I searched the forums before posting, and did get some valuable information from one recent post. [3] Searching the forum for very short keywords like "5v" and "ttl", or even "5v ttl" doesn't work well. 

 

I was going to create a small custom PCB that sits between the eval board and the inputs. The output towards the FPGA board is a 40-pin IDC connector using a standard old school hard drive cable. 

 

The input to the board will be via a single row of .1 on center male header pins. I'm going to use 6" cables from those pins to a chip clip that will connect to a DIP in another system that is primarily 5v only. 

 

The PCB will have at the center (2) 74LCX245 sold by NKC electronics. [4]. [5]. 

 

I was also going to put a ~500ohm resistor in line with the output of the transceiver towards the FPGA with the hopes of both meeting the sample circuit in the 245 datasheet and also for the series termination/minimizing reflection purposes in the 447 application note. 

 

The eval board has VCCIO locked to 3.3v and presents 3.3v on the 40-pin header, which I can use to power VCC on the LCX245. 

 

The signals are at either 7.15mhz or twice that rate, 14.3mhz. 

 

Does this sound reasonable? I'd appreciate any ideas on improving it. This is a one-off prototype design that will probably never leave my basement, but will likely be documented online. This will be my first custom PCB. 

 

Thanks, 

Keith 

 

[1] http://www.terasic.com.tw/cgi-bin/page/archive.pl?language=english&no=364 

[2] http://www.altera.com/literature/an/an447.pdf 

[3] http://www.alteraforum.com/forum/showthread.php?t=30591 

[4] http://www.nkcelectronics.com/5v-to-33v-logic-level-shifter-bo.html 

[5] http://www.fairchildsemi.com/ds/74/74lcx245.pdf
0 Kudos
19 Replies
Altera_Forum
Honored Contributor II
1,492 Views

Your design sounds reasonable. 

 

Did you notice that NKC sell pre-made boards for $1.95 

 

http://www.nkcelectronics.com/5v-to-33v-logic-level-shifter-bo.html 

 

However, I believe the DE0 has input protection clamps, so you would not need this circuit. Please review the schematic of the DE0. It should be on the CD available from Terasic. 

 

Cheers, 

Dave
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

Hi Keith, 

 

Adding a translator in between the 5V signals and the Cyclone III is a must, you wouldn't want a DC signal at the CIII input pins above 3.6V. For AC transitions you can go over that, but only for very brief periods of time, but still nowhere near 5.0V.  

 

The 500 ohm series resistor sounds a bit excessive though, usually something more in the range of 25 to 50 ohms does the trick. 500 ohms would likely give more of a saw tooth waveform with slow rising and falling edges, 50 ohms will give much sharper edges and still dampen the reflections.
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

Hi Keith, 

 

Adding a translator in between the 5V signals and the Cyclone III is a must, you wouldn't want a DC signal at the CIII input pins above 3.6V. For AC transitions you can go over that, but only for very brief periods of time, but still nowhere near 5.0V.  

 

The 500 ohm series resistor sounds a bit excessive though, usually something more in the range of 25 to 50 ohms does the trick. 500 ohms would likely give more of a saw tooth waveform with slow rising and falling edges, 50 ohms will give much sharper edges and still dampen the reflections. 

--- Quote End ---  

 

 

Usually Terasic puts protective diodes on their boards to make the I/O 5V tolerant. I know my DE2 has them, and I believe the DE1 has them as well. (BAT54S) I looked up the manual for the DE0, and it appears they forgot to include the schematic for the protective circuits, but they do mention that they exist. Might be worth a follow-up email with Terasic to verify - or just check the board for a whole bunch of 3-terminal diodes near the connectors.
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

Your design sounds reasonable. 

 

Did you notice that NKC sell pre-made boards for $1.95 

 

 

However, I believe the DE0 has input protection clamps, so you would not need this circuit. Please review the schematic of the DE0. It should be on the CD available from Terasic. 

 

Cheers, 

Dave 

--- Quote End ---  

 

 

As a matter of fact, this is exactly what I was intending on using. I linked to it originally, but because of forum policy, I couldn't show the links directly. (I posted a way to view the links) 

 

Of course, I paid $6.00 for them before they were discounted. :) 

 

I both looked on the schematic and checked the actual board. Looking near the connectors, I don't see a big multiple of anything. There are (72) free pins, so I'm guessing I would need 72 of "something" nearby. I just don't see them? 

 

Thanks 

Keith
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

Usually Terasic puts protective diodes on their boards to make the I/O 5V tolerant. I know my DE2 has them, and I believe the DE1 has them as well. (BAT54S) I looked up the manual for the DE0, and it appears they forgot to include the schematic for the protective circuits, but they do mention that they exist. Might be worth a follow-up email with Terasic to verify - or just check the board for a whole bunch of 3-terminal diodes near the connectors. 

--- Quote End ---  

 

 

I'm going to email them to double check, but I don't think they are there. 

 

I just looked at the board, and there are definitely not a bunch of 3-terminal diodes that I can see. Should there be (1) per pin to protect? aka 72 of them for my board? 

 

Thanks 

Keith
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

it appears they forgot to include the schematic for the protective circuits, but they do mention that they exist. 

--- Quote End ---  

 

Where do you see this mentioned? My simple conclusions from clamp diodes missing in the schematic is, that they have been omitted. If you have the board, you are easily able toc check, if they are present though. 

 

In my opinion, 500 ohm series resistance isn't too high for slow and medium speed signals. Another option for 5 to 3.3V conversion not yet mentioned is a simple resistive divider.
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

In my opinion, 500 ohm series resistance isn't too high for slow and medium speed signals. Another option for 5 to 3.3V conversion not yet mentioned is a simple resistive divider. 

--- Quote End ---  

 

 

I was originally going to do a resistive divider on a solderless board. But then the practicality of it hit me, 16 * 2 = 32. That's a lot of resistors, and when I sat down, it's just a big pain in the butt. I looked for alternative package of resistors, and I didn't find one that would work the way I need it. Also don't know if the resistive dividers would just naturally support the series termination requirement of AN447. 

 

Using those little boards at NKC (that contain the LCX245) would be more convenient, handle 8 connections at a time, in on one side, out on the other. Most of those solderless boards won't do two sets of ICs SIDE BY SIDE(40-pin on left, say, and the boards on right), which, for wiring purposes is the most ideal. 

 

I've wanted to do a PCB forever, and this might be a good excuse. 

 

I'd prefer soldering the connector, the two LCX245 boards, and then even 16 through-hole resistors(for termination) that have consistent lead lengths to cut, and so on. Much cleaner and permanent solution. If I get some PCB schematic capture/layout experience for a more complicated project later, then so be it! 

 

I'm pretty green with this stuff. 

 

Thanks 

Keith
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

Might be worth a follow-up email with Terasic to verify - or just check the board for a whole bunch of 3-terminal diodes near the connectors. 

--- Quote End ---  

 

 

So I did both. 

 

 

--- Quote Start ---  

There is not any protection diodes for the free GPIO pins, so we do not recommend our customer to import the 5v TTL into the board. If you have to, please try to decrease the input voltage to 3.3v to avoid damage the board, thank you!  

--- Quote End ---  

 

 

and from visual inspection, there's definitely nothing there. 

 

Thanks 

Keith 

 

0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

Well, the DE0 is intended to be a cheap board, so I guess they lopped those off - sorry for the confusion. I would suggest IDT Quickswitches over the 245 parts, though. Quickswitches have a much lower propagation time, and are inherently bi-directional - which means you don't have to reserve a signal to switch direction. They are essentially an array of FETs operating in the linear region, so they dynamically adjust their resistance with voltage. You do still get an "output enable" if needed (usually used for hot-swapping). The only downside is that they require a fairly odd voltage to operate on, 4.3V (Vccio + 1V); but as it happens, it's easily obtainable with a standard 1N4148 diode and a 10k resistor due to the low current requirements of the part. I've used them on boards in the past, and they really do a fine job. I'm using a few octal parts on a board I'm working on now, but I saw data sheets for a 20-channel part as well.

0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

The TI 74CBTD3384C  

 

http://www.ti.com/lit/ds/symlink/sn74cbtd3384c.pdf 

 

buffers used on the MAX II PCI development kit are bus-switches with the diode internal to the device. You can power them from 5V and they'll clamp bidirectional signals to 3.3V levels. 

 

I've used a bunch a bus switches on lots of designs, but never realized they had parts with the diodes internal to the devices. 

 

Cheers, 

Dave
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

The TI 74CBTD3384C  

 

http://www.ti.com/lit/ds/symlink/sn74cbtd3384c.pdf 

 

buffers used on the MAX II PCI development kit are bus-switches with the diode internal to the device. You can power them from 5V and they'll clamp bidirectional signals to 3.3V levels. 

 

I've used a bunch a bus switches on lots of designs, but never realized they had parts with the diodes internal to the devices. 

 

Cheers, 

Dave 

--- Quote End ---  

 

 

Thanks Dave for your help. 

 

I would certainly prefer something with less external parts required. 

 

I sampled these today from TI, although they only had the DBQ(SSOP/QSOP) parts available. I prefer SOIC with the larger 1.27mm pitch. I've soldered the SOIC before, and they weren't bad. The SSOP have half the pitch. 

 

Any experience hand soldering these? 

 

Thanks, 

 

Keith
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

 

I sampled these today from TI, although they only had the DBQ(SSOP/QSOP) parts available. I prefer SOIC with the larger 1.27mm pitch. I've soldered the SOIC before, and they weren't bad. The SSOP have half the pitch. 

 

Any experience hand soldering these? 

 

--- Quote End ---  

 

 

Soldering SSOP is easy. You tack the diagonal corners to get the pins aligned, and then wipe a flux pen over the pins, then solder, creating shorts by accident, then you go back with solder wick and clear the shorts. Wet the solder wick with a flux pen first, and the shorts are easily removed. 

 

Cheers, 

Dave
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

Before completely hijacking my own thread, 

 

Solder type and size? Temperature setting? Tip size and type? Solder braid size? 

 

Thanks 

Keith
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

Before completely hijacking my own thread, 

 

Solder type and size? Temperature setting? Tip size and type? Solder braid size? 

 

--- Quote End ---  

 

 

Nothing particularly special, just whatever I have lying around, lets see ... 

 

* Solder wick = PRO WICK 1801-10F 0.035" wide (i.e., narrow) 

* Small solder tip 

* Solder station set to 700F (Weller or Hakko) 

 

There's a bunch of soldering videos on YouTube, check them out to see what people do (I recall SparkFun Electronics might have some videos too, but that may be for a DIY reflow oven ...) 

 

Cheers, 

Dave
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

This video is a good example: http://www.youtube.com/watch?v=ll8gcqqxb2e. When you have more experience you can drag the tip along the pins and solder a full row in one go. 

The most important part is the flux, it makes everything so easy! You should also have a magnifying glass, and ensure that the solder tip is perfectly for a good heat conduction.
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

@jshamlet: Thanks for the info. I've definitely seen those IDT quickswitches before, but I've never used one. 

 

TI is mega fast on the samples. I ordered yesterday 1pm, this morning at 9am it was in my door. Maybe I'll dead bug one soon just to watch it work, and then design my converter PCB around it. 

 

Thanks for the link to the videos. 

 

can i attach the output of the ti chip directly to the connector going to the cyclone iii? ie without a series resistor

 

The application note (http://www.ti.com/lit/an/scda003b/scda003b.pdf) and datasheet (http://www.ti.com/lit/ds/symlink/sn74cbtd3384.pdf) doesn't seem to cover it explicitly. 

 

should i connect the fpga ground to the ground of the system i'm receiving the 5v ttl from? 

 

Thanks, 

 

Keith
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

 

can i attach the output of the ti chip directly to the connector going to the cyclone iii? ie without a series resistor

 

--- Quote End ---  

 

 

Yes. If a source termination is required, it would be needed back at the source driving the input to the bus switch. 

 

 

--- Quote Start ---  

 

should i connect the fpga ground to the ground of the system i'm receiving the 5v ttl from? 

 

--- Quote End ---  

 

 

Yes. Without it, your interface will not work. 

 

If you need an isolated interface, then Analog Devices has some nice parts; look for iCoupler. 

 

Cheers, 

Dave
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

Last, but not least, 

 

if I use pull-downs on the buffer inputs like you mention here (http://www.alteraforum.com/forum/showpost.php?p=123853&postcount=8

 

What size pull-down should I use? 

 

This will make sure that the buffer always has some valid value --- if not asserted from the device it's connected to, then at least ground from the pulldown. 

 

Thanks 

Keith
0 Kudos
Altera_Forum
Honored Contributor II
1,492 Views

 

--- Quote Start ---  

 

if I use pull-downs on the buffer inputs like you mention here (http://www.alteraforum.com/forum/showpost.php?p=123853&postcount=8

 

--- Quote End ---  

 

 

There is a big difference between using a buffer and a bus-switch. 

 

A bus-switch can be used for bi-directional clamping of signals. 

 

The bi-directional nature of this solution means that you can use the pull-ups on the FPGA I/O pins to guarantee that the I/O pins are at a valid logic level at all times. 

 

No need for pull-ups on your bus switch board. 

 

Cheers, 

Dave
0 Kudos
Reply