ADS54J60: Problem with obtaining SFDR values

Furkan Fehmi Sefiloglu

Part Number: ADS54J60
Other Parts Discussed in Thread: LMK04828

Hi Team,

Can you please check and advise with following request:

I would like to ask a few questions about TI's ADS54J60 chip, if possible. We have a problem with obtaining SFDR values which are denoted in ADS54J60 datasheet.

- We use 250 MHz ADC reference clock, which is the output of LMK04828 chip.

- SYSREF pulses are generated using LMK04828 either. We calculated the SYSREF frequency from SYSREF = LMFC / 2^N. By selecting N=4, (fs/4)/K => (1000/4)/16 = 3.90625 MHz.

- We use 4244 LMFS configuration and corresponding DDC mode & JESD link properties given in datasheet. You can find our register configurations attached (adc_config.txt).

- Fs/4 noises seem acceptable (approximately ~90dBFS).

However, our SFDR is not as fine as datasheet offers. Even if a 0 dBFS signal is applied, our maximum SFDR is around 53 dB. We always observe a harmonic spur (I think it is a second harmonic, aliased somehow) which has 2x bandwidth of original signal.

I also attached the spectrums of ADC output signal. Do you have a suggestion to improve our SFDR performance? I can provide additional information about out setup if it is required.

We have a secondary problem, last two LSB output bits of "channel A" are stuck. Depending on the input signal frequency, it is always stuck at some value ("00","01"..). Am I missing something in register configuration?

Thanks in advance

Best Regards

Furkan Sefiloglu

over 4 years ago

0 jim s over 4 years ago

TI__Guru* 91947 points

Furkan,

What tones are you testing at? The device will saturate with 0dBFs inputs. Are you using bandpass filters with the analog input? All data sheet numbers are taken at -1dBFs. Please send your register settings.

Regards,

Jim

0 Furkan Fehmi Sefiloglu over 4 years ago in reply to jim s

TI__Expert 3118 points

Hi Jim,

Please see feedback below and attachments:

- 500MHz to 1000MHz single tone frequencies are tested (Output of analog signal generator - Agilent N5171B), no bandpass filters are used with the analog input.

- Spectrum outputs for input frequencies of 650MHz, 700MHz, 750MHz are attached (-20dBFS).

Thanks in advance

Best Regards

Furkan Sefiloglu

Fullscreen 5875.adc_config.txt Download

000081
	wait 3x1msec (internal logic)
001180
005920
004F00
00200F
0021C0
0023F0
002430
002640	powerdown configs
005500	powerdown configs
400100
400200
400300
400468
60F701
	wait 3x1msec (internal logic)
400501
604EA0	Nyquist EN & Improve IL Perf
604201  Nyquist Zone A
704EA0  Nyquist EN & Improve IL Perf
704201  Nyquist Zone B
400500
604420  Digital Gain to 0 dB
605201
600001
600000
	wait 3x1msec (internal logic)
400469	
600200	ADC Test mode off
600500  Scrambling off
600709
600102	Mode 4244
603100 	Mode 4244
603200  Mode 4244
601680
40046A
601202
601602
601740
601700
400468
604100  Mode 4244
604D00  Mode 4244
607200  Mode 4244
605201  Mode 4244 & Gain Enable
600001
600000
400469
600080
60060F  K = 15 
	wait 4x1msec (internal logic)

0 jim s over 4 years ago in reply to Furkan Fehmi Sefiloglu

TI__Guru* 91947 points

650MHz input tone.pptxFurkan,

The register settings look Ok except for address 0x005500. If they want to turn off CHB this needs to be set to 0x10 not 0x00 as they have it. They need to capture data using a bandpass filter as I think this is their issue. Attached is data captured I did from home without a filter. If I had a filter, you would have seen a big improvement. Not sure how noisy their source is. The next time I get a chance to go to the lab, I will duplicate this test with a filter.

Regards,

Jim

0 Dogancan Davutoglu over 4 years ago in reply to jim s

Prodigy 20 points

Hi Jim,

I have posted this question to Furkan. Thank you for your answers.

Actually, I forgot to say, we are using a balun for the ADC input (MABA-011047. 1:1, Transmission Line Balun, 4.5 - 3000MHz). Maybe we should change it to have a max. frequency of 1000MHz.

Very appreciate your help.

Kind regards,

Dogancan

0 jim s over 4 years ago in reply to Dogancan Davutoglu

TI__Guru* 91947 points

Dogancan,

I think this transformer is ok to use. Using a transformer with a max frequency of 1000MHz will still pass higher frequencies'. They will just start to attenuate as you go up in frequency. I would try the bandpass filter first.

Regards,

Jim

0 Dogancan Davutoglu over 4 years ago in reply to jim s

Prodigy 20 points

Hi Jim,

You were right, bandpass filter has solved the problem. Now, SFDR performance matches the product specification.

Thank you!

Dogancan

Data converters

Data converters forum

ADS54J60: Problem with obtaining SFDR values