TMP422 reading 0x7F on both remote channels. Good reading on internal temp. Scope pic included

Hi Tom,

It's great that you posted the scope photos. Can you tell me what temperature readings you are getting from the device and what you would expect to see?

Can you also please send:

• Schematic and PCB layout of the device and it's connections (it doesn't have to be of your whole system)
• The Power Supply Voltage,  Digital I/O values connections (what to make sure you are not exceeding any operational rating on digital pins)
• For Remote Junction Temperature measurement, please include the remote thermal diode part number
• Also, for Remote Junction Temperature measurement, please include Series Resistance, Cdiff, and n-factor

And why are you trying different combination of diodes and swapping polarity as I don't quite understand this statement:

 I have tried all four combinations, adding one or two diodes and swapping each polarity and I see correct response of four corresponding slave addresses.

Is your scope input impedance truly 50 ohms (just checking) - it should be  more like 1Meg...

Can you show me D+ and D- and the DC levels of each.

Thanks!

Emmy,

thanks for the response.

We actually got the TMP422 working. Our issue was we did not alter the n-factor from the POR value. When we did adjust the n-factor we did read back correct temperature. What we did notice, however, was the n-factor at 20C was significantly different than 80C. Our diode is inside an ASIC we design and mfg. at IBM. We did measure diode forward bias voltage versus temperature at a constant forward bias current. The plot of V(forward_bias) versus Temperature was linear. We saw linearity at 10uA, 40uA and 100uA constant currents from 20C to 80C. We have an accurate liquid controlled thermal block on the ASIC to set the case temperature. Since the ASIC is powered off, except to the PN junction current, we are assuming the diode junction temperature is the same as the ASIC forced case temperature.

Would you expect the n-factor should be constant for a diode from 20C to 80C? Assuming we should have one n-factor, is there any factors you have experienced which would cause us to see this inaccuracy? We did a plot of n-factor which causes the read temperature to equal the force temperature versus force temperature. This function is linear, but has considerable slope. For a single diode and even multiple samples of the same type I would have expected the n-factor to be consistent over the 20C to 80C range. Do you agree? We will continue to investigate.

Hi Tom,

Thanks for the additional information. I have seen n factor variation with temperature - very rarely. It is usually constant though, so I would agree with your last statement for most cases.

One thing that looks weird in your waveforms is that it looks like the capacitance is changing with current level. Can you double check your scope to make sure it is balanced correctly and you're not having some weird saturation problems with it? It appears in your photo that at low currents it takes quite a bit longer for the voltage to settle than at high currents. That could be another source of variation. Can you check to see how the diode stray capacitance changes with temperature. If you are having settling time issues it could be disguising itself as ideality variation. We suggest you stay below 500pF total differential capacitance between D+ and D- for the TMP422. If you have a capacitor on your board you may want to remove it to see if it affects your results.

I'll check with others in my group and get back to you tomorrow if I find out anything else.


Take care,

Hello Ren,

The ASIC is 32nm. I cannot give you a sample. Sorry. This is a plot for one sample (id= B3372289_LH) of this 32nm ASIC. We used a Keithley 2602 Source Meter. We set it up to force current, measure voltage. We connected the force lines to the internal diode to forward bias and measure the voltage across the PN junction. Ploted the response for three constant force currents. We forced the case temperature of the ASIC as I stated in the previous post.

As you can see, the internal diode does a good job relating forward bias voltage at constant current with junction temperature. The diode access is a Berg stick on the ASIC test board. The ASIC is in a socket. PCB traces route the diode pins out to the Berg header. I build a little copper clad board to mount the TMP422 on. I tied this little board into the test board +3.3V and gnd planes. I placed another Berg header on this little board and wired a twisted pair to DXP and DXN. I then jumpered the Berg pins on the little TMP422 board to the ASIC test board.

I'm not sure why the TMP422 is not able to do a force current, measure voltage as the Keithley PMU can and read back the temperature data to the accuracy we saw in the plot below. I'm planning to use a TMP423 on another test board. I will design the part in so I don't need to build this little after thought board. I am a little concerned that the TMP423 may not be accurate, however, based on my results so far on the 422. Since the diode plots below look conducive I guess my little copper clad board may be the cause of the inaccuracy, but I not really confidant this is the case. So I appreciate your help. I would like to know exactly how the 422/423 forces current and reads the forward voltage. If you could explain the scope captures to me, that may help. What is going on in the 1st pluse, the 2nd, etc. and why all the toggles?

Maybe a phone conversation may be easier? Thanks for your help.

Hi Tom,

Perhaps taking this off E2E would be better as we could share more information? Please contact me and I can arrange a conference call with me and Ren.

My email is: emmy.smaragda.denton@ti.com

Thanks!