ADS54J60: ADS54J60 wrong behaviour when clipping

Patrice,

I believe this is caused by the amplifier. Can you look at the ADC input using an o-scope and probe across R17 and R25 and verify the input is either within the spec of the ADC or not?

Regards,

Jim  

Hi Jim,

It seems that the LMH6401 can drive a differential voltage greater than 1.9Vpp at R17 and R25 higher than 1.9Vpp.  In one test, I adjusted the differential voltage at R17 and R25 to 3.36Vpp.  I was expecting to see clipping (i.e. constant values of 0 or 65535 for the duration of the overvoltage) in the ADC.  However, I get ripples close to 0 and 65535.  Can you reproduce this effect on your side?

Thanks,

Patrice

Patrice,

What is the common mode voltage of this input signal? If this is not centered, this could cause the issue as well.

Regards,

Jim

Jim,

The Vcm at R17 and R25 (i.e. LMH6401) is 2.010V. The Vcm at R2 and R14 (i.e. LMH3401) is 2.138V. The Vcm is constant with or without clipping.

The datasheet mentions that the typical Vcm is 2.0V. There is no min or max.

Thanks,

Patrice

Patrice,

Can you look at the data using bits mode and let me know what you see? The display in codes mode may not be displaying properly. You can also program the device to send a fast over range bit to output on the PDN pin for CHA and the SDOUT pin for CHB. You can probe the PDN pin which is connected to GND through R31. See the data sheet for instruction s on how to set this up. There is a FOVR for CHA and one for CHB. You will also have to set the FOVR CHA ENA bit which will enable this function for both.

Regards,

Jim 

Jim,

I've captured the bits mode waveforms when there is clipping on the positive side.  I think the codes are correct.  There is no activity on the PDN pin.  ADS54J60 registers' x20, x21, x23, x24, x26, x53 and x56 (Master Page - 080h) are all 0x00.

Patrice

figure 1: Clipping on the positive side (Bits D6 to D10 are toggling)

 

Patrice,

I duplicated this testing using a TSW54J60EVM and 70MHz input tone, The results are attached. I am not sure what is going on with your setup as I show clipping at both ends when I exceed the input amplitude of the ADC input. What are using as in input source? Are you connected to the correct input SMA's?

Regards,

Jim

ADS54J60_sat_test.pptx

Jim,

I got the same results at 70MHz, 0dBm, gain =0xB. No ripple. However, if you change the frequency to 10MHz, you will get ripple.

Patrice

Patrice,

I repeated this test at 1MHz and 10MHz and the results looked like the ones I did at 70MHz. Is there a chance your source has some harmonics at this frequency?

Regards,

Jim

Jim,

I've used two different signal generators with the same results. I couldn't find see the ripples on the scope. What versions of software/firmware/configuration are you using? I'll try to reproduce your setup.

On my side, I was using the following setup:

High Speed Data Converter Pro v4.50
FPGA image : ADS54J60_LMF_8224

ADS54JXX EVM GUI v1.6
C:\Program Files (x86)\Texas Instruments\ADS54Jxx EVM GUI\Configuration Files\LMK_Config_Onboard_983p04_MSPS.cfg
C:\Program Files (x86)\Texas Instruments\ADS54Jxx EVM GUI\Configuration Files\ADS54J60_LMF_8224.cfg

Patrice

Patrice,

These are the files I am using and the same versions of GUI's. Did you press the reset switch on the ADC EVM after loading the LMK config file? If so, you may have a bad EVM. Send me your address and a phone number and I will send a replacement TSW54J60EVM.

Regards,

Jim 

Jim,

I've tried to press the reset switch on the ADC EVM after loading the LMK config file but it didn't help.  I've tried also a third generator with the same results on both channels.  At first, it looks the data is clipping correctly but when you zoom in there is some ripple on either the positive, the negative or both sides...  At this point, I would like to try a replacement board.

My address is: 

Gage Applied

attn : Patrice Plante

1 Rue Provost, Lachine, QC H8S 4H2

Lachine, QC

H8S 4H2

Canada

Thanks,

Patrice

Patrice,

Did you order this board through a local distributor? If so, I think they are the ones that can handle getting you a replacement board. Let me know if you have any problems with them getting you a new board.

Regards,

Jim

Jim,

Could you help us to get our replacement board from Arrow? We have send back our board on March 3rd but Arrow's RMA group cannot give us any status update. Did Arrow contact TI about the replacement board (i.e. Arrow RMA #QM3959770)?

Thanks,

Patrice

Patrice,

Send me your address and I will send you a board.

Regards,

Jim

Include a phone number as well.

Jim,

My email address is patrice.plante@gage-applied.com and my phone number is 514-633-7007 x3046.

Thanks,

Patrice

Patrice,

I will need a mailing address, not an email address.

Jim

Jim,

Here is my mailing address:

Gage Applied
attn : Patrice Plante
1 Rue Provost
Lachine, QC
H8S 4H2
Canada

Thanks,

Patrice

Jim,

I've received the replacement evaluation board from Arrow.  Thank again for your help on this matter!

I've injected a 10MHz signal, 0-dBm, gain =0xB.  When the signal is clipping, I don't get constant values of 0x0 or 0xFFFF.  It looks like a voltage ripple effect.  See snapshots below.  Can you see this ripple on your board?

Patrice

Patrice,

I am checking with the design team regarding this. This is not an operational  mode we would suggest a customer to use. Can you explain why you are doing this? Would the use of the OVR feature help you instead of monitoring saturated data?

Regards,

Jim

Patrice,

What do you notice if you increase the input amplitude by 5dBm?

Regards,

Jim

Hi Jim,

If I increase the amplitude by 5dB, the clipping lasts for a longer period, but the ripple has the sample amplitude.

For our application, we might occasionally saturate the inputs of the ADC.  With other TI ADCs, the amplitude would clip at 0x0 or 0xFFFF and there is no ripple.  If you can reproduce this effect, I would like to confirm the source of the ripple (analog front-end or ADC internal circuitry).  If we can fix it by modifying the AFE or changing a register in the ADC that would be great.  If not, I simply will need document this limitation for our customers.

Patrice

Patrice,

If you have the gain setting to 0x0B, this is a gain of -9.2dBm and you cannot saturate the ADC with 0dBm input. Did you use a gain of 0xB0? I tried this value and did not show any ripple when I had the ADC saturated. See page 57 of the data sheet for the gain calculation used by register 0x44. If you are using a gain of 0x0B, the input required to saturate the ADC will be larger than is specified by the input spec of this part.

Regards,

Jim  

Jim,

Sorry for the confusion, I was setting the gain at 0x0B in the LHM6401. I'm always using the default value (0x0) in register x44.

Thanks,

Patrice

Patrice,

Base on your post, I was assuming you were using the ADS54J60EVM. Since I now know you are using the TSW54J60EVM, I am wondering if your problem is related to the amplifier. Do you see this same thing on both channels with the input saturated? Do you see any sign of this ripple on the ADC input pins? Like I mentioned in the previous post, I did not see this on the ADS54J60EVM.

Regards,

Jim  

Jim,

I've just tried the experiment on my new replacement board and I'm seeing this on both channels.

Thanks,

Patrice

Patrice,

Any sign of the ripple on the ADC input pins?

Regards,

Jim 

Jim,

No sign of saturation and ripple on ADC inputs pin of both channels (R17, R25 or R2, R14).

Thanks,

Patrice

Patrice,

I do not remember if I asked for the config files you are using. Can you please send these? Are you using a filter on the IF input?

Regards,

Jim

Jim,

I'm using the following original configuration files (see zip file in attachment):

- ADS54J60_LMF_8224.cfg

- LMK_Config_Onboard_983p04_MSPS.cfg

I'm not using an external filter on the IF input.

Thanks,

Patrice

cfg_files.zip

Patrice,

I was able to duplicate your issue. If I then added digital gain inside the ADC, I was able to get more of a result that you would expect. I ran this test on both the TSW54J60EVM and the ADS54J60EVM. The data is attached. I will send this to the design team to get their feedback regarding this.

Regards,

Jim

TSW54J60_IF_10M_983p04M_Fs.pptx

Jim,

The DC offset correction is a hidden register in the ADS54J60.  Can you provide the register address and field description?

Thanks,

Patrice

Patrice,

The two attached files will allow you turn off the DC correction and turn it back on.

Regards,

Jim

 ADS54J60_LMF_8224_DC_ON.cfgADS54J60_LMF_8224_DC_OFF.cfg

Hi Jim,

I am working with Patrice on the same project. I notice that the digital gain (reg 0x6044/0x6052) will fix the clipping issue. Does the digital gain affect the ENOB?