5V to 3.3V bus and vice versa

[Altera_Forum]

MC68000 to CycloneII 

 There is something wrong with quickswich devices. As I know MC68000 cant support 3.3V (some versions can but mine just cant). When i use quickswich any type i only get 5V to 3.3V translation and no vice versa. I need to use dual voltage level translator with auto sensing. Only parts that I can found with this is txs0108e even sn74lvc8t245 have direction control. For last two days I m reading about Pericom articles and those articles are just like IDT have for quickswich part. Devices are working from higher to lower and I m thinking that I just cant use those devices. For some devices who work with 5V can drive lower voltage levels that is ok but on my situation is not. For example PI5CX and PI3CX series are only related to 5V to 3.3V translation no bidirectional and then i need to use bus enable input like on most devices. From A to B bus or from B to A bus, that is not practical. Also some Pericom devices PI74LVC4245 who have two voltage on each bus side but no auto sensing again I need to set direction sides. I just dont know, as much i m reading only thing is txs0108e that I m using on design right now. Also using Pericom or IDT devices we can use pull up resistor connected to higher power source. I just dont know....

[Altera_Forum]

Just to clarify; the problem is that the MC68000 input threshold voltages are 5V HCMOS compatible, but not 3.3V compatible? 

 You could use two buffers; a bus switch between the MC68000 bus, so that the Cyclone II device sees 3.3V signals, and then a 5V buffer from the Cyclone II to the MC68000. When the MC68000 performs a bus access that corresponds to the Cyclone II, the bus switch will route the control signals from the MC68000 to the FPGA, and then the Cyclone II can drive the response via the 5V buffer (which would have 3.3V compatible inputs).  Cheers, Dave

[Altera_Forum]

Yes that is the problem. I know about solution that you proposed but it is to complicated. I must find one device who can perform all of that. I just dont have the space for that on the board. But thank you for your response.

[Altera_Forum]

--- Quote Start ---  Yes that is the problem. I know about solution that you proposed but it is to complicated. I must find one device who can perform all of that. I just dont have the space for that on the board. But thank you for your response.  --- Quote End ---    How much logic do you need in the FPGA, and what is it doing?  If you do not need too much logic in the FPGA, then use a 5V device. The older generation FLEX10KA FPGAs are 5V tolerant, and are still available (I have boards that use them). What about a MAX CPLD?  Cheers, Dave

[Altera_Forum]

I use about 60% of logic elements of EP2C5T144C so i cant use anything below that.

[Altera_Forum]

Are the signals going from the Cyclone II to the MC68000 'fast'? 

 If they are 'slow', then can you consider putting pull-ups to 5V on the signals (bus pull-ups)?  The Cyclone II device can then drive high (to Voh ~ 3.3V), and then tri-state the output driver, the pull-up can then take the signal over the last few volts to 5V.  5V I/O clamps can then be used on the Cyclone II I/O to protect it from the 5V signals (see the DE2 board GPIOs).  Cheers, Dave

[Altera_Forum]

Hm thank you I ll check that, but basic idea is to cyclone II take over the mc68000 bus, and it should be fastest as it can be. PI5C34X245 is the device I would like to use, because it has 32 bus and for replacing of MC i could use only two of those devices. In Pericom manual says that PI5C is 5v=>3.3v but no vice versa. I just dont understand why quickswich devices cant work <=> their names says that they can.

[Altera_Forum]

What about these TI parts? 

 http://www.ti.com/lit/ds/scds305/scds305.pdf  I didn't look, but there are probably 16-bit versions.  Cheers, Dave

[Altera_Forum]

This one is the same it has direction controle. Lower to higher and higher to lower. I contacted TI and they provide me solutions. Thank you for your time.

[Altera_Forum]

--- Quote Start ---  This one is the same it has direction control.   --- Quote End ---    Eh? No it doesn't, read the first line of the datasheet  "Provides Bidirectional Voltage Translation With No Direction Control Required"    --- Quote Start ---   I contacted TI and they provide me solutions.  --- Quote End ---    Great! I'm glad you found something.  Cheers, Dave

[Altera_Forum]

Yes yes I read all but no first line. .) Thank you I might think about using something like that but only need to find 16 bit or more but based on this device I ll look right now. I think that this version is the old version of txs or txb series and becouse of that they could have 16 bit versions of the device. Txs and txb series does not have more then 8 bit devices.

[Altera_Forum]

Here's TI's translator guide, p3 has the parts that auto-sense direction: 

 http://www.ti.com/lit/ml/scyb018d/scyb018d.pdf  Looks like 8-bits is the maximum width supported by TIs devices.  Cheers, Dave

[Altera_Forum]

Yes but there are some old devices (hard to find now) who could capable of doing translations. This is all ok but i ll have also to try with PI5C34X245 from pericom also vith PI7C series and with QS32x245 from IDT. Also there is trusted QS3384. I ll get prototype PCB in 20 or so days and i ll report here my investigation. I think that this part of integration of old to new devices is not investigated, or we common people cant find informations. I was thinking that maybe this problem is simple one and noone take time to investigate it. Also production of devices goes up and they just dont want to interface new with old. Also voltages is going down, one day we might talk about interfacing 0.1V to 5V old system... As much I m reading about this for over past months I m thinking about that any bus translation device can work fine if you tune it in some other way, just we need to think about PCB space, and cost of production...

[Altera_Forum]

Unless you are trying to upgrade a legacy product, while trying to preserve a large software, or adapter board market, its unlikely that you would deliberately pair incompatible logic standards such as your MC68000 and a new generation FPGA. 

 As you first pointed out, there are MC68000 devices with LVCMOS compatible logic interfaces, so I would expect most people to select that if they absolutely had to use an MC68000. Although, I would expect a new design to leverage ColdFire or something more closely compatible with the MC68000.  In my designs, the FPGAs are used for signal processing. With a new generation FPGA, I can process significantly more data, and generally this means I have to transport more 'results' off the FPGAs. This in turn requires faster external buses, network interfaces, and hence a faster processor.  Rather than reusing a specific processor, I try and update to the latest processor with a compatible instruction set, eg., in my case PowerPC processors from Freescale.  Cheers, Dave

[Altera_Forum]

Yes that is what I was talking about. But for me problem is that I just cant replace MC68000 becouse they did not produced coldfire that can do that. It just cant emulate all the instructions and some are not the same. Maybe one day that I can manage to do some things about that.

[Altera_Forum]

--- Quote Start ---  But for me problem is that I just cant replace MC68000 becouse they did not produced coldfire that can do that. It just cant emulate all the instructions and some are not the same. Maybe one day that I can manage to do some things about that.  --- Quote End ---    Yeah, I can understand. Good luck with your design.  Cheers, Dave

[Altera_Forum]

These clamping devices need low ohmic pull-up to achieve sufficient speed at the high voltage side. This may be a problem with the drive strength of MC68000. But if the bus speed requirements aren't too high, it should work. 

 In addition, I don't understand, why you can't generate a direction signal from the MC68000 bus control signals? Any kind of bus extenders does it.

[Altera_Forum]

I decide to go with txs0108 becouse it has internal pull ups