TLC3578: Some analog inputs are off by fixed amount, others are very accurate.

Charles Kaseff

Part Number: TLC3578
Other Parts Discussed in Thread: ADS8678

I am using a TI TLC3578IPW to measure power supply voltages on my board. Some of the digital results are very accurate compared to the analog voltages on the ADC input pins. Some of the digital results are off by a fixed amount. I have taken many readings, and the digital results are very consistent. In other words, the results do not vary more than 10mV with every read.

The worst-case reading is for a 1.0V supply from a TPS7A9001DSKR LDO. It reads 0.936V where the analog voltage is 0.999V. The digital results for the 5V and 3.3V supplies are within 20mV of the respective analog voltages.

CFR is programmed to 0xE04 (BTC, Short Sampling, Internal Conversion clock, Single-ended, one-shot, EOC)

SCLK is 5MHz frequency. Vref is 4.0V with 20mV ripple p-p.

Why is the digital reading for the 1V supply much more off compared to the other supplies? I can supply schematic to TI.

9 months ago

0 Tom Hendrick 9 months ago

TI__Guru**** 174226 points

Hi Charles,

Please do share your schematic, we'll take a look.

0 Charles Kaseff 9 months ago in reply to Tom Hendrick

Prodigy 65 points

Hi Tom,

How do I upload my schematic to TI. This is company IP, so I don't want to post it to a public forum.

Also, I read Test registers B, C, D and got the following values:

B: 0

C: 0x2002

D: 0x1ff9

0 Tom Hendrick 9 months ago in reply to Charles Kaseff

TI__Guru**** 174226 points

Hi Charles,

You can send me a direct message - send a friend request and we can chat offline. Also, just wanted to make you aware that the TLC3578 is going on 'last time buy', so if you are committed to long term availability, you might consider changing your design to use the ADS8678. For the test codes, are you in binary two's compliment or straight binary mode? What reference voltage are you using?

0 Tom Hendrick 9 months ago in reply to Charles Kaseff

TI__Guru**** 174226 points

Hi Charles,

For the test voltages, please refer to the table on page 19, the test voltages should be as follows:

Test C is FS-, B is mid-scale and D is FS+. Please review this video which may help explain the errors you are seeing with your conversion results:

https://www.ti.com/video/5476574757001

0 Charles Kaseff 8 months ago in reply to Tom Hendrick

Prodigy 65 points

The value that I read from the ADC for Test Reg B (Midscale code) matches the value 0000h or 0v. The value read for Reg C is 2002h, which is 2 count off of 2000h. The value read for Reg D is 1ff9h, which is 6 count off of 1fffh. Do these values explain the 63mV error that I am reading on my 1V analog supply? If so, what can be done to improve the values besides the 10uF and 0.1uF caps between REFM and REFP, which are already on the board next to those pins?

0 Tom Hendrick 8 months ago in reply to Charles Kaseff

TI__Guru**** 174226 points

Hi Charles,

The deviation in the test voltage does not match. If your reference is 4.096V, your LSB size is 4.096/16384, or ~250uV. The 6LSB delta would only be ~1.5mV. Did you watch that video I suggested to you?

Because of the Thanksgiving holiday in the U.S., TI E2E™ design support forum responses may be delayed from November 25 through December 2. Thank you for your patience.

Data converters

Data converters forum

TLC3578: Some analog inputs are off by fixed amount, others are very accurate.