Exact match. Not showing close matches.
'[EE]: Level translator fun'
Has anyone here been using the the Texas Instruments TXB0104 (and other
members of that product line, e.g. TXB0108)?
I've designed one into a GPS controller board, and I'm seeing some really
odd behaviour. The GPS chipset feeds the TXB like this:
TXD o----[220R]------| TXB0104 |-------> IOCOM_RXD
RXD o----[220R]------| |-------< IOCOM_TXD
In this diagram, "Side A" is on the left, and powered by a National
Semiconductor LP5951MF-1.8V (1.8V LDO). "Side B" is on the right, powered by a
USB port. Both Vcc's are bypassed with 100nF ceramics within millimetres of
the chip, grounding straight to the topside ground plane. The 5V input
additionally has a 4u7 multilayer ceramic near the input to the 3.3V voltage
regulator for the GPS chipset.
IOCOM_RXD and TXD go straight to an FTDI FT232R USB-to-RS232 controller
(actually an FTDI TTL-232R cable). Drive strength on those is about 20mA, and
drive strength on the GPS side is similar.
As far as I can tell, both sides are operating within spec per Vcca/Vccb, but
the TXB isn't translating properly. If I put a scope probe on TXD (before the
resistor), I can see the NMEA data blitting across the serial link. If I put
it after the resistor (on the "A" side of the TXB), I see a constant low. If I
scope the output of the TXB0104 (B-side), I see what appears to be 0V/5V noise.
AIUI, my GPS chip is driving with 20mA at 1.8V. Ohm's Law says the series
resistor will bring that down to (worst case) 8mA. Even so, the TXB0104
datasheet states that it should work correctly as long as the driving device
has a drive capacity of at least 2mA each way.... Hmm...
I'd really appreciate a sanity-check here...
Just looking quickly at the datasheet.
This is a bidirectional device. Is it possible the IOCOM_RXD signal is
forcing the input (TXD) to ground?
What happens if the OE pin is pulled low - does the TXD signal at the
input pin now appear correctly on the scope?
2009/6/4 Philip Pemberton <philpem.me.uk>:piclist
|Richard Prosser wrote:
> Just looking quickly at the datasheet.
> This is a bidirectional device. Is it possible the IOCOM_RXD signal is
> forcing the input (TXD) to ground?
It shouldn't be possible -- the FT232R's RXD line is input-only, and the
pullup resistors aren't fitted on the TTL-232R cable.
> What happens if the OE pin is pulled low - does the TXD signal at the
> input pin now appear correctly on the scope?
A number of things happen:
- TXD at the input pin of the TXB0104 becomes recognisable. That is to say,
the noise disappears, and the data stream looks like a normal 4800bd NMEA
stream (although I can't tell what it's sending because neither my scope nor
the logic analyser can speak RS232). So it's probably fair to say the GPS
chipset is working properly.
- The +5V and +1.8V lines become much cleaner (the noise level falls
dramatically). I suspect this is a side effect of the TXB0104 outputs no
I'm going to see if swapping the chip does any good (though I need to fire up
the hot-air station to do so), but at this point I'm starting to wonder if the
TXB0104 is a turkey... Occam's Razor suggests that probably isn't the case,
but bgatliff's replies in this thread on AVRfreaks are making me wonder:
Just to reiterate: I'm seeing noise on the 5V lines, but NOT when the TXB0104
is disabled. 1.8V is showing 50mV pk-pk, 5V (the one coming straight from the
USB chipset) is showing 58mV pk-pk. Turn the TXB0104 on (OE connected straight
to VCCA) and the whole thing goes to hell (>250mV pk-pk noise on both +1.8V
|Philip Pemberton wrote:
> I'm going to see if swapping the chip does any good (though I need to fire up
> the hot-air station to do so), but at this point I'm starting to wonder if the
> TXB0104 is a turkey... Occam's Razor suggests that probably isn't the case,
> but bgatliff's replies in this thread on AVRfreaks are making me wonder:
OK, swapping the chip did nothing, and neither did piggybacking a pair of 10nF
ceramic capacitors on top of the 100nF power decouplers for the TXB0104. So
whatever's causing this, it isn't noise on the power lines.
Also, the "noise" isn't actually noise, it's the chip dropping into some sort
of self-oscillation. Between 40.75 and 41.6MHz if anyone's interested. Exactly
identical on both chips.
Interestingly this is only happening on the I/Os that are being fed to the
TTL-232R cable. There are two pins left open on the I/O connector, one of
which is wired to the TXB0104.
So it looks like the TXB0104 doesn't like driving another chip via a chunk of
wire, but has no such problems driving an open circuit...
Now I just need to see about hunting down a mating connector for the GPS
board, a 74LS or 74HC buffer chip, and a chunk of Veroboard...
The TXB0104 can't drive wire. Adding a 74LS244 to buffer the TTL from the TXB'
has stabilised the circuit. The noise on the 5V rail is gone, as is the noise
on the data stream. The GPS chip is talking to the PC, and is currently
tracking 8 satellites (not bad considering the antenna is a cheap little 5V
Trimble mag-mount patch antenna).
And now onto the fun part of the project: syncing a 10MHz oscillator to the
GPS 1PPS signal.
Alan B. Pearce
>And now onto the fun part of the project: syncing a 10MHz
>oscillator to the GPS 1PPS signal.
Oh, you mean like this (uses a PIC too!)
He has a PDF of the original QST article. I have been going to build one of
these into a thermos flask to make a temperature stabilised GPS disciplined
oscillator, for some time. One day a tuit will happen ... ;)))
More... (looser matching)
- Last day of these posts
- In 2009
, 2010 only
- New search...