3.3v to 5v Bidirectional logic level shift

logic level shifter

im working on a project where i need to use a 2.5" harddrive
in a device with compact flash (true ATA) slot.

while the interface between the harddrive and CF controller in True ATA mode
is fairly straight forward. as seen by the 100's of CF-IDE adapters on ebay.
the problems start when going the "other way" when the CF controller is 3.3v logic and the harddrive is 5v logic.

so i need to do some bidirectional level shift on the ATA logic.

have been looking at a couple "level changers" but so far i have
only found unidirectional chips " 5v input A to 3.3v output B"
not " 5v I/O to 3.3v I/O

How do i make such a bidirectional levelshift, that is also fast enough
for ATA logic, data transfer ?

any ideas, suggested chips / circuits
anything would be appreciated.

/Kyndal

3.3v to 5v level shifter

the problems start when going the "other way" when the CF controller is 3.3v logic and the harddrive is 5v logic

Click to expand...

Most today's PC boards are using 3.3V IO to drive PATA disks with 5V IO, cause no modern IO chip set has 5V IO. They are also used with CF cards in true IDE mode with commercially available CF bridges (actually passive adapters). Some manufacturers supply main boards with CF on-board slot, they also use 3.3V IO.

However, CF spec requires voltages above 3.3V on some pins, while ATA doesn't. As far as I know, this isn't a real issue with (most) existing CF drives. It may be rather a problem, that true IDE mode isn't operating reliable with some CF products. It may be the case, that CF cards from a manufacturer known as working correctly suddenly fail in a new series (cause the controller chip was changed).

My first suggestion is to use no level translation at all or at least not for bidrectional signals (the latter are O.K. with 3.3V also by CF spec). If you feel a need to do anyway, e.g. to improve noise margin, dual supply bus buffers (e.g. from TI) should be OK. They need a direction control from other bus signals, however.

3.3 to 5v level shifter

thanks for the quick response.

i know that CF cards on a pc motherboard is not a problem,
as the cards can do both 5v and 3.3v communication.
and i have used this many times in the past too.

but just to clarify this, im going the "other way" than conventional
adapters and using a harddrive on a Camera!
not a CF card as a harddrive


so if i understand correctly you are saying
that the 3.3v logic from the CF HOST (camera) should be sufficient to drive the
5v input logic on a 2.5" hard drive. ??

And that the 5v logic outputs from the hard drive, wont hurt the 3.3v logic in
the CF Host ?

id hate to trail an error on this one, as it is some very expensive equipment


if this is the case
(that the 3.3v is suficient logic to drive the harddrive.)
i could just add a bunch of 3.3v or 3.5v zenar diodes
to drop the 5v on the logic lines down ?? leaving 3.3v signals un touched?
never tried zenars in this kind of application.

would this work?


any idea?




/Kyndal

5v 3.3v level shifter

I did understand the unusual setup at first sight.

that the 3.3v logic from the CF HOST (camera) should be sufficient to drive the
5v input logic on a 2.5" hard drive. ??

And that the 5v logic outputs from the hard drive, wont hurt the 3.3v logic in
the CF Host ?

Click to expand...

The first guaranteed by ATA spec, the second - not necessarily. The said PC IDE controllers are 5V tolerant, but I don't know with your camera. You may use series resistors (50 to 100 ohms) and schottky diodes to 3.3V supply as voltage clamp. Or a 5 volt tolerant bidirectional buffer (1 74LVC16245 or 2 74LVC245).

As another point. Are sure, the controller is using true IDE and 16 bit mode?

level shifter 3.3v 5v

on the voltage clamp,
will zener diodes be fast enough for ATA aplication?

Camera.
lets just say that this camera is not the "runn of the mill" camera
we shoot very large RAW images, Continuously in a Aerial photography application using a Hasselblad house with a digital back.
and the data firewire port is used for Shutter sync, so unavailable for data..

the camera can use microdrives, and right now is using large udma CF cards
but they are not available big enough, so they have to be a changed a couple of times during each flight. Which is NOT optimal.
and yes these cards are BIG but with 64gb cards soon to be available
this problem might solve it self..

But for now, sadly there is no Commercial products available that can do this.
so i have been asked to look into a solution..

But while the 74LVC16245 looks good, it needs to get a "direction" signal
which i suppose i might be able to get this from the data read/write strobe signals or something

i like the lowtech zener solution better.. so far...

/Kyndal

3.3v hard drive

Most controllers have clamping diodes to the power rail as part of the I/O structure. All that is usually necessary is a series resistor to limit the current. There used to be logic chips that I believe were called quickswitch. They contained just pass MOSFETs and were advertised as bi-directional logic translators. The reason that they worked as is their on resistance was high enough to limit the current.

You can check for the diodes in the controller of the camera with just the diode check function of a multimeter with the camera un-powered. Test from an IO pin to the camera's power rail. If the diodes are not present, you can add schottky diodes as already mentioned. Zeners are not a good bet here. I think they are too slow and the capacitance is too high. The schottky diodes are available in packages with a diode array that reduces the total parts count.

5v to 3.3v level shifter

okay
thanks for the replys guys..

i defiantly have some ideas for experimentation

going to contact the manufacturer of the digital back. about the controller details
who knows. it might just be 5v "tolerent"

/Kyndal