TMDS442: TMDS442 HPD Source Current Limit?

Hi,

So with the same video decoder, the intermittent video issue does not show up with the pattern generator, but does show up with the SBC? 

What is the voltage on HPD1 when there is an intermittent video issue? Can you also replace the 953ohm resistor with a 0ohm to see if it helps solving the issue?

TMDS442 datasheet list the IOH and IOL test current for VOH and VOL. Even though they don't indicate a limit for TMDS442, these current ratings still give a good idea of the maximum recommended output current as shown below

Are you also seeing voltage drop on VCC when there is an intermittent video issue?

Thanks

David

Hello David,

Thank you very much for looking into this issue we are experiencing.

1) So with the same video decoder, the intermittent video issue does not show up with the pattern generator, but does show up with the SBC? 

Yes, the failure are more frequent with the SBC.  Failures had been seen with other sources, but they were less common and the video could could be recovered by toggling the output's input source.  This is something we started doing when the decoder doesn't detect video.  However, this wasn't effective with the SBC.

2) What is the voltage on HPD1 when there is an intermittent video issue? 

The voltages remained consistent with what was shown in the table whether it was in a failed state or not.  Video appeared to be streaming through the TMDS442 part in a failure state from what we could see on a scope, but the video decoder couldn't get a DE lock.

 3) Can you also replace the 953ohm resistor with a 0 ohm to see if it helps solving the issue?

We did experiment with different values for the series resistor, but saw video failures when a 0 Ohm was used.  Decreasing the resistor value lowered the HPD High and HPD Low voltage values.  The customer stated that there is a 10K Ohm pull down on the SBC and because of this the HPD output will be divided. I drew a simply equivalent circuit below to help explain.

When the HPD1 is high from the TMDS442 there is a divider from 5V between R901 and R1 and the 3.3V HPD1 output is divided by R743 and R1 causing the HPD to drop to 3.8V instead of 5V.  And when the HPD1 is low the 5V is divided between R901 and (R1 + R743) causing the voltage to only drop to 2.3V.  We found using a 500 ohm series resistor (at R743) dropped the high to 3.6V and the low to 1.58V.  Reducing the resistance could help get the low voltage down further and we may experiment with this, but it doesn't seem to affect the SBC operation.

4) TMDS442 datasheet list the IOH and IOL test current for VOH and VOL. Even though they don't indicate a limit for TMDS442, these current ratings still give a good idea of the maximum recommended output current as shown below

I did see the 4 mA IOH and IOL note, but wanted a confirmation of this because of the behavior observed with the output.

5) Are you also seeing voltage drop on VCC when there is an intermittent video issue?

No, I asked our EET and she said the VCC was consistent.  She is double checking this again and I will provide her findings.

Something else that I should add is that we did perform a leakage test on a number of parts and saw that they were failing this test.

Thank you again and I look forward to your feedback.

Hi,

Per the HDMI spec, the VIH at the source is min of 2V, and max of 5.3V. VIL at the source is min of 0V and max of 0.8V. The TMDS442 HPD1 being a 3.3V will meet both the VIH and VIL condition. But it appears from the description, that pulldown resistor on SBC create a voltage divide network, and the strong 10k resistor is pulling the HPD VIH below the SBC source VIH level. But with the R901 1k resistor, the voltage on HPD is being pulled back up and above the SBC source VIH level and solved the intermittent video issue. 

If without the R901 1k, and just tuning the R743, are you able to solve this intermittent video issue? My concern with the 1k pulled up to 5V is whether the HPD1 can still be properly driven low.

Thanks

David

Hi David,

I agree that the low voltage is out of spec like it is.  We will have to experiment without using the 1K pull up to see if we can get the video to reliably work.  The HPD voltage did drop to 0V in this configuration.

We haven't been able to explain why the 1k pull up makes the video work reliably.  I was hoping that maybe you would see a problem with the measurements that I provided. Without the 1K pup the HPD voltage was a bit higher with the pattern generator than the SBC.  Why do you think this is?

Thank you again for your feedback.

Hi,

Do you know the HPD VIH inside the SBC? I am thinking with that 10k pulldown, the voltage on HPD without the 1k pullup might not be meeting the HPD VIH threshold inside the SBC. But with the 1k, the voltage is being pulled up above the threshold and then it works. 

For the video generator, do we know if the 10k is also present on HPD? I am thinking either 10k is not present, or a weaker pulldown resistor, or a lower HPD VIH might explain why the video generator works.

Thanks

David

Hello David,

We don't have access to the SBC schematics to analyze it.  However, we did experiment with using a bench supply to supply the HPD instead of the TMDS442.  HPD from our display was disconnected from the SBC at the external cable.

From this testing we found that the SBC requires the HPD to be above 1.7V.  Video stopped switching on when the HPD voltage dropped below that.

Our very first test failed and it had the R901 1K PUP resistor removed.

1) Turn SBC On

2) Wait a few seconds

3) Turn Display On

4) Wait a few seconds

5) Turn bench supply HPD = 2.8V

In this failed state the video didn't recover by toggling the HPD on and off.  The voltage was increased up to 5V and this also didn't make it recover.  Video displayed after resetting the SBC.

When the SBC boots and the video displays successfully the HPD could be used to turn the video off and then back on.

We then added the PUP back with the HPD still disconnected.  Failures were only seen when cold spray was used on the video board.  Admittedly the TMDS442 part is only rated to 0C so maybe the cold spray is too much.  We will continue testing tomorrow when the board is back at ambient temperature.

Do you have any more thoughts on why the pull-up could be needed?  It seemed odd in this case considering there was no load.

Testing this way at least proved the low HPD voltage level was not a problem and neither was having a load on the HPD.

Hey Matthew, 

Based off of the data you have collected the voltage of the HPD isn't the issue. It could be that adding the R901 pullup increases the amount of current that can be sourced to HPD. Is there a way you can measure the current at the SBC with and without the R901 pullup. 

Hello Vishesh,

I though the SBC was too much of a load for the HPD1 driver as well.  However, this last test had the HPD1 from our display disconnected from the SBC and failures still occurred without the PUP.

My first post list current measurements from the HPD1 with and without the PUP.  Please refer to "1K Ohm PUP to +5V + SBC" and "1K Ohm PUP removed + SBC".  Was this what you were looking for?

Hey Matthew,

Yes it is! I should have checked dither entire thread before responding. The currents you are drawing should be within spec.

I have a small suggestion. In your 5 step test procedure try turning on the HDP supply before turning the display on.  It might be an order of operations issue. As sometimes when the display turns on it drives the HPD to 0V and then pulls back up. This process may be interrupted when you turn HPD on after the display is already on. 

Also am I correct in understanding that the display is failing only when the display is cold? and if so what temperature is cold?

No worries Vishesh, this is a lot of information to go through and I appreciate your feedback.

You may be right that the HPD needing to start high.  Whether the SBC or our display requires this  I am not sure.

With a 1K PUP at R901 and a 953 ohm series resistor at R743 the HPD voltage always stays above 2.3V and is the configuration we stopped seeing failures.  We experimented with lower series resistor values and started seeing them when the LOW state voltage dropped below 0.8V (i.e. HDMI low threshold).  For example, with a 100 ohm series resistor the HIGH state voltage was 3.47V and the LOW state voltage was 0.63V.  The display would therefore actually be setting the signal low when it is turned on.

Our display's software can control the HPD once the MCU is running.  Usually the HPD would be held low from a POR while the AFE is initialized.  This is done because the EDID is stored in non-volatile memory and this prevents the source from reading it before it is ready.  After the EDID is loaded the HPD is allowed to go HIGH or LOW with the 5V input.  We can maybe change the timing of this.  We did find the SBC doesn't actually require a EDID.  In a test I left it blank and the SBC displayed video.

Failures were at ambient temperature after our display sat powered off.  Once the display ran for a while it would always detect and display video.

Sounds good. While you're getting the MCU running, can you clarify on the block diagram of the circuit setup?

1) is the SBC the source, and AFE the display/sink?

2) If so the HPD should be going from the AFE to the TMDS442, is this how you have it setup?

I am also not fully understanding the issue. Does the pull resistor added fix the issue of the display not working properly until it runs for a while? or does the issue still exist with the pull up resistor?

Sorry for asking so many questions but I want to fully grasp the problem before I can try to solve it :)

Hello Vishesh,  I'm getting ready to fly out of state and will be unavailable next week.  Our senior electrical engineer Maria Fey is going be available to answer questions while I'm out.  We have been working together on this so she will be very knowledgeable.

1) Yes, the SBC is the source and sink is the the TMDS442 -> AFE.

2) Yes. The SBC is connected to the TMDS442 HPD1 and the TMDS442 HPD sink is connected to the AFE.

Adding the PUP so the HPD voltage is always high allows the video to not fail at ambient.  Without it the display needs to run for 30 seconds or so before the video will work.  I can't explain this either, but is what we observe.

Would it be possible for us to send our video board schematic to you for review? My simplified circuit lacks some details and this should make things easier to understand.

Thank you again for your Support!

Hi Matthew

Would you please accept friendship request so you can send us the schematic in a private message?

Thanks

David