Logic outputs to logic inputs of devices with different Vcc

[Tom48]

Hello engineers! :)I have a question bothering me i quite can't figure out.

I have a Priority encoder:

http://www.digikey.com/product-search/en?vendor=0&keywords=ls148d&stock=1

datasheet:

http://www.ti.com/lit/ds/symlink/sn74ls148.pdfAnd this RF-Switch:

http://www.digikey.com/product-detail/en/ADG904BRUZ/ADG904BRUZ-ND/1007291

datasheet:

http://www.analog.com/media/en/technical-documentation/data-sheets/ADG904_904R.pdfNow the problem is that the RF-Multiplexer's supply voltage is 2.75V while the supply voltage of the Priority encoder is 5V.

I now would like to connect the Logic output pins (A0,A1) of the priority encoder to the logic input pins (A0,A1) of the rf-mutiplexer.

I assum that i can't just connect these 2 devices like that, thus is need to decrease the voltage somehow.I checked the datasheet of the priority encoder (page 8), it says there is a High-level output voltage of 3,3V typically (at Vcc=Min and High-level input voltage = 2V).

The first question is: If my High-level input voltage is 5V and not 2V like under test conditions, will also my High-level output voltage increase here?

Im just not sure what voltage i will get here on the output so i hope there is someone who can give me a hint here.The second question is, how do i decrease my voltage according to my needs here.

The RF-switch has a typicall data delay of 10-12ns, the priority encoder about 15ns.

So, i need a quite fast logic processing as well.I thought about a voltage divider in the first place but had to find out that this method might be just way to slow according my needs.Another possibility could be a level shifter IC, in case you would recommend this option, are there some you would use for this matter?I appreciate every little hint!
Thanks a lot in advance.Kind regards, Tom.

 

[meBigGuy]

You don't want to overdrive the mux input signals from 5V logic unless they are specified as being 5V tolerant. (and they are not, Absolute max, Inputs to GND is –0.5 V to VDD + 0.3 V

One possibility is to use low voltage cmos logic. But, you are probably coming from a 5V system, so somewhere you will need level shifters.

There are a number of possibilities. GTL may be the fastest in the high to low direction, but it's kind of weird ( http://www.nxp.com/documents/application_note/AN10145.pdf ).
I only use it when I need bidirectional level shifting and can deal with the pullups in one direction. (only good solution I've ever found for multivoltage I2C)

This page has "normal" level shifters.
http://www.ti.com/logic/docs/transl...5.0&searchDirection=1&voltageOut=2.5#voltintf

Like the SN74LVC1T45

 

[Tom48]

You don't want to overdrive the mux input signals from 5V logic unless they are specified as being 5V tolerant. (and they are not, Absolute max, Inputs to GND is –0.5 V to VDD + 0.3 V

One possibility is to use low voltage cmos logic. But, you are probably coming from a 5V system, so somewhere you will need level shifters.

There are a number of possibilities. GTL may be the fastest in the high to low direction, but it's kind of weird ( http://www.nxp.com/documents/application_note/AN10145.pdf ).
I only use it when I need bidirectional level shifting and can deal with the pullups in one direction. (only good solution I've ever found for multivoltage I2C)

This page has "normal" level shifters.
http://www.ti.com/logic/docs/transl...5.0&searchDirection=1&voltageOut=2.5#voltintf

Like the SN74LVC1T45

Thank you so much for your reply, already helped a lot! :)

So just a voltage divider made out of 2 resistors would not be acceptable at all for my undertaking here, right?

As i never had to work with level shifters before, I am a little confused.
Lets take the SN74LVC1T45 as an example here:
There are 2 pins for 2 different supply voltages. (Vcc_a and Vcc_b).
Lets assume i would like to shift a logic level from the Priority encoder output (priority encoder has a supply voltage of 5V)
to the logic level input of the mux (the mux has a supply voltage of 2.7V).
So a shift from High to Low supply voltage.

So in this case i had to power the Vcc_a pin of the SN74LVC1T45 with 5V and the Vcc_b pin with 2.7V?
Then power the DIR pin with a "high", to get the A data to B bus direction.
After, connect the logic output of the priority encoder to the A pin of the SN74LVC1T45 and the B pin to the logic inputs of the mux, did i get it right?

Thank you in advance, i really appreciate your help! :)Kind regards, tom.

 

[meBigGuy]

1. The voltage divider would work, but has the issues (like speed) you initially mentioned. It would be the simplest in terms of components, but needs careful design from the point of view of the range of chip output voltages over process (what output voltage at what load current varies over production characteristics and from vendor to vendor)
You may be able to use a 5V CMOS encoder with more drive (especially better "high" drive than LS logic).

2. You could also use open drain NMOS or open collector NPN transistors with pullups to 2.7V, which are also slow. But that approach is simple, especially with 5V drive.

3. What you described with the SN74LVC1T45 seemed correct. If it is not, it is very close.

 

[Tom48]

1. The voltage divider would work, but has the issues (like speed) you initially mentioned. It would be the simplest in terms of components, but needs careful design from the point of view of the range of chip output voltages over process (what output voltage at what load current varies over production characteristics and from vendor to vendor)
You may be able to use a 5V CMOS encoder with more drive (especially better "high" drive than LS logic).

2. You could also use open drain NMOS or open collector NPN transistors with pullups to 2.7V, which are also slow. But that approach is simple, especially with 5V drive.

3. What you described with the SN74LVC1T45 seemed correct. If it is not, it is very close.

Thanks a lot! :)

Do you know a "better" or faster Priority Encoder?
As i searched a lot and i just was not able to find smth. notably faster than the one i have mentioned in the first post.

Just to make sure: In case i use the SN74LVC1T45, when i power the Vcc_b with 2,7V, am i safe with that for sure?
My down shifted output will never exceed the voltage on the B output pin of the SN74LVC1T45 that i set it to, using the Vcc_b pin?
That would be totally awesome! :)thank you so much! :)

 

[meBigGuy]

There are many faster TTL logic families than LS (like S, AS or ALS). There are also fast low voltage CMOS families. This is TI's listing. Other companies have logic families also.
http://www.ti.com/lit/sg/sdyu001aa/sdyu001aa.pdf

I don't know what you driving the 148 with, and whether you could do the whole thing in 2.7V logic, or level shift somewhere else, etc.

The level shifter will work as you described. But, use the SN74LVC2T45 or other parts with more inputs/outputs.

 

[Tom48]

There are many faster TTL logic families than LS (like S, AS or ALS). There are also fast low voltage CMOS families. This is TI's listing. Other companies have logic families also.
http://www.ti.com/lit/sg/sdyu001aa/sdyu001aa.pdf

I don't know what you driving the 148 with, and whether you could do the whole thing in 2.7V logic, or level shift somewhere else, etc.

The level shifter will work as you described. But, use the SN74LVC2T45 or other parts with more inputs/outputs.

Thanks a lot for your help, i really appreciate it! :)

I use an ADC-flash structure before the 148 Priority Encoder, made out of a simple voltage divider and 3 high speed comparators (also supply voltage of 5V).
This is the comparator that is included in the adc flash (right after the voltage divider):
http://www.digikey.de/product-detail/de/TLV3501AIDBVT/296-18248-1-ND/806961
datasheet:
http://www.ti.com/lit/ds/symlink/tlv3501.pdf

Do you see a problem with this combination?

thank you! :)

BTW, what do you mean by your last sentence? (would this level shifter fit into this circuit or what do you mean?)
"But, use the SN74LVC2T45 or other parts with more inputs/outputs."

 

[meBigGuy]

Here is another possible solution. Turns out all LVC logic has 5V tolerant inputs and works at 2.7 volts. But, no priority encoder.
http://www.nxp.com/documents/application_note/AN240.pdf
http://www.nxp.com/products/logic/family/LVC/

 

[meBigGuy]

You are converting 2 bits, so use a 2 bit part. Maybe that was already your plan.

 

[Tom48]

You are converting 2 bits, so use a 2 bit part. Maybe that was already your plan.

Ah, ok, yes I am aware of that, thanks! :)

Just curious about this idea:

Lets say i take the comparator i posted above (http://www.digikey.de/product-detail/de/TLV3501AIDBVT/296-18248-1-ND/806961)
and put it on a 2,7V supply.
Then i connect the negative comparator input to the 2,7V rail as well.
The positive comparator imput, i would then connect to the adc-flash structure that is ahead, so it would be driven by the same comparator (but that comparator that is on a 5V supply and a part of the adc structure). IN that case a "high" could be accomplished and as it is an I/O rail to rail comparator the 2.7V comparator that is driven by the adc 5V comparator should "level shift" the signal to 2.7V as its on a 2.7V supply, or am i mistaken?

Do you get my idea?
If not i might create a schematic fast.thank you! :)

 

[meBigGuy]

Nice try, using the comparator as a level shifter.

If you look at the "Absolute Max ratings" for the TLV3501AIDBVT you see that the inputs are limited to VDD + 0.3V. There is a note saying that if that is exceeded you have to limit the current to less than 10ma, which implies you can add a series resistor and drive it from 5V. So, with that, it will work I guess.

The maximum input specification is generally due to the ESD protection structures that clamp the input to VDD with a diode. When that diode conducts it can turn on a parasitic 4 layer device that shorts VDD to VSS and destroys the part (called latchup). It generally takes a fair amount of current (maybe 40ma) to cause latchup, so they say limit it to 10ma on that particular part.

The reason I mention this is that there is a similar diode in the RF switch to clamp the input (it has the same VDD + 0.3V rating). So, in theory, you can simply add a series resistor to limit the current and drive it from 5V. I hesitated to mention that initially since it is a questionable practice and there was no current limit note in the ADI spec sheet. Also, in some mixed signal parts you can get strange side effects when you overdrive the inputs.

The combination of a load resistor directly on the LS148 output to load the weak output driver (say 1K or smaller) and a series 1K resistor to the RF switch input may work fine . (or may not)

 

[Tom48]

I just learned a LOT, thank you so much!
I see, many possibilities to accomplish the goal with each of them coming with up and downsides... (or even more uncertainness ;) )

As speed is important to me ill check out S, AS or ALS families to see if i can find a faster priority encoder (the ls148 has a delay of 15ns i wasnt able to find smth. faster yet).
I guess you can not name a faster one off the top of your head :)
Do you know about a faster comparator than the TLV3501AIDBVT?
Also was not able to find smth. faster yet but to make sure i did not overlook smth. important i like to ask ;)

Thank you so much, you really helped me a lot! :)
About the level shifter:

I wanted to know if the idea with the comps would be possible or just completely dump.
I guess i will go with the 2bit level shifter offered by TI.
Seems pretty fast to me, very secure and i would need only one more additional device (the SN74LVC1T45 itself)
Or do you completely disaggree with that? :)

 

[meBigGuy]

I wanted to know if the idea with the comps would be possible or just completely dump.

It's a good idea, and would work with the series input resistor. Don't know about the speed. Putting a tiny charge dump cap across the resistor would gain some speed.

74F148 may be the fastest encoder in std logic. Way faster than LS.

You can use 5V tolerent buffers as level shifters.
VHC and AHC and similar families can do level translation since they have 5V tolerant inputs, but are pretty slow at low voltages.
http://www.nxp.com/products/logic/family/AHC_T/#overview

http://digsys.upc.es/sed/ED/unitats/unitat_1_6/1_3_8_TI_Digital_Logic_Families.pdf (OVT in the right column means overvoltage tolerant)
http://www.ti.com/lit/sg/sdyu001aa/sdyu001aa.pdf

ECL is the fastest logic, but is way way different than "normal" logic. Many times the ECL/TTL conversion uses up all the speed gain.
There are faster comparators with PECL outputs. Depends on what lengths you need to go to for speed.
http://www.ti.com/product/lmh7324

You would have to read up on ECL and figure out how to priority encode and interface to your switch. I won't be able to help with that.

 

[Tom48]

It's a good idea, and would work with the series input resistor. Don't know about the speed. Putting a tiny charge dump cap across the resistor would gain some speed.

74F148 may be the fastest encoder in std logic. Way faster than LS.

You can use 5V tolerent buffers as level shifters.
VHC and AHC and similar families can do level translation since they have 5V tolerant inputs, but are pretty slow at low voltages.
http://www.nxp.com/products/logic/family/AHC_T/#overview

http://digsys.upc.es/sed/ED/unitats/unitat_1_6/1_3_8_TI_Digital_Logic_Families.pdf (OVT in the right column means overvoltage tolerant)
http://www.ti.com/lit/sg/sdyu001aa/sdyu001aa.pdf

ECL is the fastest logic, but is way way different than "normal" logic. Many times the ECL/TTL conversion uses up all the speed gain.
There are faster comparators with PECL outputs. Depends on what lengths you need to go to for speed.
http://www.ti.com/product/lmh7324

You would have to read up on ECL and figure out how to priority encode and interface to your switch. I won't be able to help with that.

ok, thank you! :)

I came across the 74F148 already but it seems pretty hard to get. (in case you know some more id be glad to check them out)
Thanks a lot, i will check the fast families out soon (mentioned in the pdfs).
I do not intend to use ECL logic, but thanks for pointing that one out! :)

So, actually i should be good with the level shifter by TI you suggested above right? (SN74LVC1T45)
Seems pretty fast and realiable to me.

Thanks a bunch for all the info! :)