FT230X 3V3OUT pin connected to 3.6V, Safe?

The FTDI ft230X datasheet states that, for the output pin of the internal LDO:

When VCC is 3V3; pin 8 is an input pin and should be connected to pin 10.
page 7.

Meaning that 3V3OUT of the internal LDO should be connected to the power supply (VCC) when the chip is being powered by 3.3V.

The datasheet also states that the maximum output of the LDO is 3.6V. page 17.

The 3V3OUT is internally connected to some modules(USB transceiver and reset generator). page 13.

I am assuming that since the LDO "may" output 3.6V, that the USB transceiver and reset generator will be fine if connect 3.6V to both the 3V3OUT and VCC pin.

Is this assumption correct?

This is for an application where an input voltage of both 3.3V and 3.6V are possible on VCC.

Clarification:

Ordinarily, when VCC is at 3.6V the LDO will be attempting to regulate down to 3.3V. BUT! since the LDO output is ALSO connected to VCC, there will be no current through the LDO. The LDO may attempt to lower it's output voltage by increasing it's pass element's resistance, but this will have no affect since the output is help at 3.6V.

Is this safe? Having the same voltage on both the input and output of the LDO, that is higher than what the LDO is trying to regulate down to?

(we know that VCCIO, USB transceiver and reset generator can handle 3.6V, so lets forget about that for now)

Good assumptions. But no. You missed one part of the datasheet. Section 5.3 shows:

• 3V3
• 3.3v regulator output
• Min 2.97 > Typical 3.3 > Max 3.63 in Volts
• VCC must be greater than 3V3 otherwise 3V3OUT is an input which must be driven with 3.3V

The 3.3v LDO is a Low Drop Out regulator. The datasheet does not specifically state how much of a drop voltage is required (How much greater VCC needs to be than 3v3). At 3.6v VCC, that would give 0.3v of a drop out voltage, which is well within most LDO minimum range.

Emailing FTDI would be your best option though, as they are always welcoming and helpful on design issues.

Also, page 7 is for the QFN version of the chip, with different pinout from the SSOP version. Double check which you have and are referencing to, as the pinouts are not same.

Edit: As Oli has pointed out, there is a Errata for the FT230x that shows an issue with the 3v3 LDO in the Rev. B chips. This does not affect Rev A, or Rev C (latest). For both packages, you can tell the Revision by:

The date code format is YYXX where XX = 2 digit week number, YY = 2 digit year number. This is followed by the revision number (A/B/C).

• The OP is not using the internal regulator though, since it's a direct 3.3V input in this mode if connected to VCC (otherwise 3V3OUT becomes an input which must be driven with 3.3V). I assume since they say it becomes an input, the regulator is bypassed (so there is no drop) I must admit the datasheet is not at all clear on this matter though. – Oli Glaser Feb 18 '13 at 2:17
• @OliGlaser That is only if VCC is 3.3v. If it is at 3.6v as the OP is driving it, then it should still be an output with the LDO regulating down the 3.6v to 3.3v, as noted in Section 5.3. Even Section 3, the pinout description, notes "If VCC is 3v3". – Passerby Feb 18 '13 at 2:25
• Yes, yhat's the bit I'm not sure about - there is no level range mentioned (I can see) at which the input switches to an output. I think there has to be some range/hysteresis, since otherwise if the supply varies a little it would be switching from input to output. For example, is it an input or output at 3.35V? I'm wondering about this and the LDO drop -if it was 4.3V in Rev B, what is it in Rev C? – Oli Glaser Feb 18 '13 at 2:40
• @MrRadiotron It's either or. If the LDO is regulating, do not tie it's output to VCC. If your VCC is at 3.6v, the LDO should regulate fine (unless it's Revision B silicon, then it won't). You could test it with a 2 leds in parallel at 20ma each (40ma out of the 50ma limit for the 3v3LDO), when powering VCC from 3.6v. – Passerby Feb 18 '13 at 2:56
• @MrRadiotron - Good point - even if the LDO was trying to regulate, then since a standard LDO cannot sink current, it should be safe yes, although since we have no knowledge of the inner setup we can't be sure. It's output voltage should just sit at 3.6V. It wouldn't be the desired way to do things though anyway. – Oli Glaser Feb 18 '13 at 2:56

Since it's just within specs, I would say this is not ideal, but probably okay, yes (edit - see discussion, I think Passerby has a good answer and points, so I would contact support, or simply make sure your supply is either 3.3V or 5V). As long as you can be absolutely sure your supply never exceeds 3.6V. Bear in mind 3.6V is an absolute maximum rating, which you should try and avoid running near if at all possible.
According to the datasheet though, it confuses matters by showing a self powered example with a VCC ranging from 3.3V-5.25V, but not connected to 3V3OUT.

However, this is a mistake, since according to the errata note:

This makes sense since any linear regulator needs at least a slightly higher voltage (usually at least 200mV, for LDOs) to maintain regulation.
I'd definitely e-mail the support to make sure, and point out their datasheet is unclear on what is an important issue.

• Oli, the errata is only for Rev B chips. Rev A and Rev C (Newest) do not have that issue. – Passerby Feb 18 '13 at 2:17
• @Passerby - Yes I noticed that. I'm wondering how the datasheet can still show the incorrect example schematic though, since the 3V3OUT obviously cannot regulate with the range shown in the latest datasheet (unless they are assuming the internal drop is okay, which isn't mentioned as far as I can see). Even if it's not 4.3V, it has to be over the 3.3V shown. – Oli Glaser Feb 18 '13 at 2:27
• @Passerby - If you look at this FT232 datasheet on pg.53 it shows a self powered 3.3V option with VREGIN, VCC and VCCIO all connected together. I'm wondering if this IC has the same kind of design but with the output option on the same pin also. However it's meant to work though, I'd be contacting support, since it's confusing :-) – Oli Glaser Feb 18 '13 at 2:32
• Yea, that chip is more explicit. VCCD is an Output when 5v VREGIN is used, and an Input when 3.3v VREGIN is used. Section 5.2 does show VREGIN at 5v is typical 5v, but Ranges from 3.6v to 5.5v. VRegin is considered 3.3v from 3.3v to 3.6v. But I agree, it is very confusing. – Passerby Feb 18 '13 at 3:29

From FT230X datasheet, section 4.2, paragraph "+3.3V LDO Regulator":

The +3.3V LDO regulator generates the +3.3V reference voltage for driving the USB transceiver cell output buffers.

From FT230X datasheet, section 5.3, table column "3V3":

3.3v regulator output: min=2.97 V, typ=3.3 V, max=3.63 V.

Condition: VCC must be greater than 3V3 otherwise 3V3OUT is an input which must be driven with 3.3V

There is no explicitly defined limits for the transceiver, thus we (could) presume they the same as the regulator output limits, i.e. (3.3 ± 0.33) V, or 3.3 V nominal ± 10 % tolerance.

Therefore, if your power source can achieve ±0.03 V (i.e. less than ±1 %) regulation tolerance (in all working conditions, including the temperature range as Kuba right noticed out), it can feed the IC safely. If it cannot, the feeding scheme is unsafe.

Also note that USB 2.0 Specification, Sectiom 7.3.2 says the transceiver voltage has a 3.3 V nominal with a 3.6 V maximum for low- and full-speed devices.

Are you saying that you have a supply that has 3.6V output with 0% positive tolerance over the intended temperature and input voltage range? Have you actually tested a batch of your supplies to ensure the 3.6V limit is not broken? What control do you have over that voltage? If this is an input into your device, then I'd consider it very unlikely that the real supplies used by your customers have non-zero positive tolerance.

So, it has the smell of bad engineering. I would never do it unless on a one-off where I can hand-select the parts and do a burn-in test. For volumes higher than two (yes, two), I'd never do it.

• I don't see how this is an answer to the question in any way. Perhaps it would be best to make this a comment? – Anindo Ghosh Jan 3 '14 at 18:37
• ? Your question was whether it's safe. I've enumerated what unlikely conditions would need to be fulfilled in order for it to be safe. The hopefully clear answer is: no, it's not safe, unless your conditions are tightly controlled. What else is there to say? – Kuba Ober Jan 3 '14 at 18:53