I'm looking to translate from/to 1.8v and 2.5v for a RGMII interface running at an effective rate of 250MHz. It looks like the only solution is a TVC or CB3T with strong pullups. It looks like the CB3T uses less power, but i don't understand the TVC spec on Ipass current. Is this per FET? what is it? i've checked out the app note(scea035a), but i'm a little nervous using these parts. Does the FET stay on in the high to low direction when driven both high and low (so no need for strong or any pull-ups)? When driving from low to high voltage, does the output track the input on a rising edge until the FET turns off(at Vref), then the pull-up takes it the rest of the way to 2.5v (in my case)? so the rise time only has to go another .7v or so to be a valid signal level for the 2.5v device? i'm trying to estimate the real rise time hit due to the pullup.
I've also got to translate SPI and other signals, am looking for a solution for these too (25MHZ). In looking at the spec for the, avc8t245, they don't really spec the rise and fall time, but seem to add it to the prop delay. Am i correct in this assumption? I could use the same part mentioned above (CB3T), and just back way off on the pullups to cut power use. i looked at the TBX0106, but the prop delay and rise/fall times sort of kill you at max (7+5+12ns). That's not going to run at the claimed 50Mhz...
Hi,
For the AVC8T245, right, the prop delay includes some rising time and falling time, and this part should be no problem working at 25MHZ.
For the TXB0106, under different voltage power supply , it can support different max datarate, for 50MHZ(100MBPS), the voltage should be VccA>=2.5v while VccB>=3.3V. you can double check it in the Page 7 in the datasheet.
For your TVC and CB3T device quesiton, I will forward to the related person to answer it. any further thing, please let me know.
BR
Junjie
Hi Bob
CB3T devices are designed to translate down to Vcc. They will require a pullup when translating back up. Ipass is max current alowed through the switch per FET. I am fairly sure you can not get to 250Mhz using CB3T or TVC. If you use a resister that gives you fast enough edges you probably will not have the current to drive it. The CB3T actually has a 100Mhz spec but I could not find a max frequency spec for TVC.
We have some buffers that can run this fast. You could use the AUC245 which would translate down to 1.8V and when going the other way it would still be 1.8V but that should be good enough Vih for a 2.5V part. We also have thes in smaller bit widths.
I don't think the AUC will quite do it. my Vih on the up voltage is 2.0. The Voh min at the low voltage chip is 1.17 min. The AUC Voh max would be 1,8 if vcc is 1.8, that's not good enough, and if vcc is 2.5v, the Vih is 1.7, too high for the 1.17v coming in. Why don't you think the CB3T or TVC won't work? The pullup resistor would need to be around 150 ohm to 2.5v, giving a current of about 17ma. The driver when going low should be able to handle this (24ma rated). And the state going high is determined by RC of 150*10pf-1.5ns. That's in the ballpark (.63 of risetime). Am i missing something?
I don’t know for sure. Here is what I am going by. See apps info at end of datasheet.
PCA9306
The PCA9306 is a TVC die.
i think they're talking about a lot more C load. I2c goes up to 400pF. i'm looking at 5-10pf. the question is how the FET behaves as it's switching on/off. Can you dig up anybody that's tried this? Thanks.