Digital inputs bandwidth

I have not yet tested the digital inputs (probably by lack of digital probes :wink: )

But I have 2 comments:

  1. Looking at the schematics I am a bit surprised by the values found on the input circuit.
    The 11K resistors combined with the capacitance of the ESD protection diode (which is 110pF) makes a large lowpass filter. It is a pity as there are ESD diode that have very low pF capacitance at very low price.
    This will surely lower digital inputs performance.

  2. As far as I understand it, the inputs of the FPGA are configured in 2,5V input. The ESD protection diodes clamp voltage to 5V (7.3V max during transient). Don’t you think the inputs will be stressed if used on 5V logic (well perhaps not as the current is limited by the resistor)?
    On the Quad’s specifications, the voltage level for the digital inputs was not specified. It could be good to specify it before people burn their Quads.

Greetings all,

After examining the Quad schematic and FPGA datasheet, I found out that pins 24 and 25 on the FPGA are used for Chan C and Chan D, respectively. As it happens, those pins are in what the datasheet calls “bank 3” (page 49 of datasheet). Bank 3 inputs have an absolute max rating of 3.6V (page 91 of datasheet). Only banks 0 thru 2 have a 5V tolerance mode if you set VCCIO to 3.3V. This feature is not available for bank 3 (page 10 of datasheet).

I recommend not exceeding 3.3V on Chans C or D until a mod can be installed, possibly installing lower voltage ESD diodes at the inputs.
Absolute Max Ratings circled.jpg
VQ100 footprint circled.jpg

please help to state how did you find the diode is 110pF? i do not find the related statement in the datasheet . maybe your suggestion can help the designer a lot .

I have found the doc here : http://www.cj-elec.com/files/pdf/Mlkj_20093514583430.pdf

…does anybody know WHICH VERSION of the hardware are affected by this bug:

  • 2.2
  • 2.6
  • 2.7B

Is somebody from Seeedstudio reading?

Thanks
Patrick

Thank you very much for your inspiration. After a disappointing test of the digital input I found this thread. (HW Version 2.7B)

I replaced the bespoken diode with a PESD3V3U1UT (just 12 Cent).
It’s a 3.3 V ESD diode with just 1pf!
Datasheet:
nxp.com/documents/data_sheet … UT_SER.pdf

Look at these examples of an external 500 Khz signal:
original diode:

after the replacement:

Of course: Absolutely no warranty!

Stuemper

Hello stuemper,

The digital channel response improvement is great.

(Edit. Removed dumb frequency question. My mind was taking a break and didn’t let the rest of me know.)

Thank you.
Jerry

Pay attention!

PESD5V0 (NXP): ESD protection diode + compensation diode
RClamp0502ba (SEMTECH): 2 ESD protection diodes

PESD5V0 - SOT23
RClamp0502ba - SOT523 (it isn’t so convenient, as sot23)

Here my experiments with RClamp0502.
IMAG011.png

Hi alvadep,

Thank you, you are absolutely right! I have just tested channel C and have overseen that not both diodes are ESD. The dealer didn’t mention that and I read actually only page 1 of the data sheet :frowning:
Your warning must be repeated: with my diode there is no more ESD for one of the digital channels!

What do you think about this one?
(TVS-Diode array WE-TVS 2-Ch+VDD 3,3 V )
docs-europe.electrocomponents.co … 07e9ea.pdf
(Has to be mounted rotated by 90°)

Can you explain: Why have most of the ESD diodes an input capacitance of 100 pF and others (like the diode in my first try) just 1 pf?

Frank

Quite good option.
There can be some difficulties at installation.
Equivalent: http://www.semtech.com/circuit-protection/low-capacitance/sr3.3/

I would recommend RClamp0502: http://www.semtech.com/circuit-protection/low-capacitance/rclamp0502ba/
Low Capacitance equivalent CESD5V0.
And here is how RClamp0502 is established:
DSC00200_3.JPG

Thank you alvadep. I will try to get the recommended diode.

Frank

It would be not bad to try TVS diodes array in SOT-143 (SEMTECH SR3.3 or similar). The capacitance is a little more, but more reliable protection. In my opinion even it is not bad possible to establish with the minimum alterations in board (PCB).

I was thinking of doing this replacement but I can’t find a source for the RCLAMP0502 in UK.

What about TPD2EUSB30, which is designed for USB3 protection?

ti.com/lit/ds/symlink/tpd2eusb30.pdf

Package is a bit smaller which would be my main concern.

I’ve just spent a bit of time looking for something suitable and my conclusion is:

a) 3.3V
b) SOT23 (or anything similar shape)
c) Low capacitance

Pick 2.

The only part that’s anywhere near is the RClamp0502 but that’s 5V, it’s the wrong size and I have no idea where to get one.

So…I was thinking: How about a hybrid solution?

Put a PESD3V3U1UT where the existing diode is, it gives 3.3V clamping on both channels plus ESD protection on one of them. Then add a 2-pin ESD protection diode across the input connector on the other channel (where the connector is soldered to the board). It doesn’t matter really what the voltage is for this diode so it’s no problem finding something with single picofarad capacitance (or less).

Any thoughts?

Using two devices sounds a bit messy but if you try it let us know how it goes.

My thoughts on the USB3 device is it is very low capacitance and actually has a superior protection strategy in using ordinary diodes combined with a clamp diode so that the higher capacitance of clamp diodes is effectively shielded by the low capacitance ordinary diodes.

It is in a smaller pack which may make it a bit fiddly to fit. I’m not overly concerned about any differences in protection voltage as the series limiting resistor will lower fault current levels that could possibly get to the actual device before the main protection kicks in.

I’ve ordered one to give it a go.

The problem is that the protection diode is on the same side as the series resistor.

Consider for example if there is 10V input voltage. Then the FPGA diodes will start clamping at 3V (koti.kapsi.fi/jpa/stuff/pix/dso_ … urrent.png). Because the voltage is clamped at 3V, absolutely no current will flow to the 5V protection diode. Such a diode is completely useless, it won’t do anything until the FPGA input is already broken.

I’m not sure if 3.3V protection diodes would be effective either. The FPGA input diode forward voltage will probably rise a bit when the current rises, so it might still help something.

Optimal solution might be replacing the R41 and R42 with a series connection of two resistors, with the clamp diode in between them. But this is probably quite difficult to do.

Yes. You’re right that the input resistor configuration doesn’t allow for best protection here.

I thought though that the existing external diodes were also higher stand-off than internal and so it would not be making anything worse.

I’m no expert but I think that main risk is during ESD transients where the situation is quite complex with the transient conditions being significantly different from what a simple stable low voltage analysis indicate. So for sustained low voltage overload the internal will be taking the current but the series resistance will be fairly effective in limiting energy. Obviously putting on medium voltages for sustained periods is not going to do the series resistor much good depending on its rating but that is a low risk when using these inputs for what they are intended. For high voltage pulse the external diodes may help redirect the energy depending on the transient behaviour of the internals and provide better protection

If the external diodes have high capacitance, they ruin the digital input bandwidth. Because the digital inputs otherwise go directly to FPGA, they should be good for up to 72MSps without the external diode.

The TPD2EUSB30 pack I have ordered is designed for protecting 6GB/s USB 3 connections and has 0.7pf capacitance to ground. It should have negligible impact on achieved bandwidth.