Hello,
Trioflex, could you explain me how you could measure the 1ppm THD of an oscillator with with an ADC (AD7760) given for -105dB (5.6ppm) ?...
You can use a stepped passive attenuator to see how the THD of your ADC
move xith input level.
To measure very low level of THD you will need to use a twin-tee notch filter, reducing drastically the level of signal. Then, you ADC will produce no (or very low) spurious and you show the spectrum of the residual THD of the oscillator.
(This has been already fully explain on the thread in my signature).
FRex
Trioflex, could you explain me how you could measure the 1ppm THD of an oscillator with with an ADC (AD7760) given for -105dB (5.6ppm) ?...
You can use a stepped passive attenuator to see how the THD of your ADC
move xith input level.
To measure very low level of THD you will need to use a twin-tee notch filter, reducing drastically the level of signal. Then, you ADC will produce no (or very low) spurious and you show the spectrum of the residual THD of the oscillator.
(This has been already fully explain on the thread in my signature).
FRex
Well if you look at my FFT plots then you see that H2 and H3 are at least 20dB over the ADC noise floor, hence when those spurs would disappear measured THD I think should already be at or below 1 ppm. This with direct measurement and no notch filter used. Or put it that way, AD7760 if used with best performance configuration should be able to get a little bit better results compared to what we have. We are however already very close to the figures and FFT plots that ADI is giving in the datasheet.
Please understand this: I am testing the ADC not the oscillator, I am assuming as of all measurements done by different 3rd parties, that is if operated properly Victors oscillator has H2 and H3 below the levels I am measuring right now (if ADC and diff amplifier are good to maximum IC specs).
2 Frex: you do not happen to have any 10KHz LT super oscillator PCB's overleft?
Please understand this: I am testing the ADC not the oscillator, I am assuming as of all measurements done by different 3rd parties, that is if operated properly Victors oscillator has H2 and H3 below the levels I am measuring right now (if ADC and diff amplifier are good to maximum IC specs).
2 Frex: you do not happen to have any 10KHz LT super oscillator PCB's overleft?
Looks great. I'm already committed to Victor's stuff though. I just ordered a Par-Metal series 14 box (RFI proof to -70dB).
Hm, I did buy some candy from the market for 7EUR.
Placing Victors oscillator in the candy box (BOX not connected to common ground!) removed all spurs of 50 and 100Hz from the spectrum.
Will be drilling holes tomorrow to assemble the generators properly
If it's a steel box then it will shield EMI regardless of grounding. If you want RFI shielding also, then the box has to be grounded. That's my simple minded understanding anyway.
I suppose we could put poor Victor's oscillator into all kinds of tins, like they do with headphone amps... coffee cans, Altoids tins, Earl Grey Tea tins... etc. Maybe some exotic Estonian snuff tin.... <g>
What i am saying is if these spurs come from the ADC itself, they never disappear.
You seem to build an interesting project that allowing to perform spectral
analysis from DC to more than MHz. This type of tool could be very useful for many hobbyist. Is a commercial product ? Do you think to share it in this forum ? Or maybe sell it ? I would be interested to know more about it.
Regards.
Frex.
Hi
we are working on commercial projects. What I can say that is it likely that we will have also products that we will be offering for sale.
In the picture above is our current VICTIM, AD7760 SMD module, visible with part of the base board. The module is 22.5x22.5 size large, it fits PLCC68 SMD socket footprint. SMD pads are 1.27 mm pitch.
On the module are implemented AD7760 with ALL needed bypassing, ADP7102 low noise LDO's, 5V clock buffer IC, 4.096V Voltage Reference. For operation the module should be powered from 7.5V supply (analog part) and 2.5V for digital part.
AD7760 datasheet says also that it can be used with Audio standard sampling frequencies, so I assume it has been used for some audio use
We have several more analog blocks in similar form factor, and we plan more of them.
Right now we measured 0.01% THD and 79dB SNR when generating 100KHz sinus with our AD9952 based signal generator module, this without bandpass filtering, only 1MHz lowpass. I must make more homework, but I think 0.01% THD at 100KHz is already not bad at all.
I promise to stick around
Ah, and I will be assembling frex 10KHz oscillator so I will be able to compare these two oscillators with AD7760.
Ah, the AD7760 module is being tested with USB test board, that is not so good, we did not know that FTDI FT2232 fast streaming only works in port A! So we can not do continuous streaming at full 2.5MSPS rate
High speed data acq stuff is hard. I think the best accessible examples are the Pico Tech scopes. They seem to have very good software and are able to get pretty deep memory stuff working.
The USB stuff is full of serious headbangers when you get near the limits. XMOS may be the best option since they have as many as 8 cores in their chips at 500 MHz. Their tech staff can be very helpful with this kind of stuff. You can also do useful preprocessing before sending the data over to the host.
.01% THD at 100 KHz is quite good.
There are some other exceptional superdacs. These: LTC2209 - 16-Bit, 160Msps ADC - Linear Technology are pretty remarkable, and pretty expensive. I think it the device used in this gadget TimePod 5330A which I would love to be able to afford. Its also a high speed USB device. You may be able to get some good ideas from his implementations.
The USB stuff is full of serious headbangers when you get near the limits. XMOS may be the best option since they have as many as 8 cores in their chips at 500 MHz. Their tech staff can be very helpful with this kind of stuff. You can also do useful preprocessing before sending the data over to the host.
.01% THD at 100 KHz is quite good.
There are some other exceptional superdacs. These: LTC2209 - 16-Bit, 160Msps ADC - Linear Technology are pretty remarkable, and pretty expensive. I think it the device used in this gadget TimePod 5330A which I would love to be able to afford. Its also a high speed USB device. You may be able to get some good ideas from his implementations.
Hi dirkwright,
It called a header and a header connector would mate.
There is all sorts.
David.
.025" square posts on .1" centers and he uses the first and third position on a three pin connector. Start here for the pins: 929400-01-36-RK 3M | 929400E-01-36-ND | DigiKey they list compatible connectors at the bottom but there are many others
Timepod uses LTC2216 and opal kelly FPGA module. We have bit heavier gear here, PCIe board with LTC2208 and Virtex 5 FPGA capable of full bandwidth dual channel streaming from LTC2208.
It's all the software and know how what makes the value for the tools.
Timepod looks really interesting, we are having trouble with jitter measurements as well.. or better to say we can not measure it properly. But we need usually to measure higher frequencies. Still the timepod is nice piece of hardware.
Or just twist the wire around like you did, add some solder on top, and call it "MIL-STD"
Really
Well, maybe so. I looked up the parts kindly mentioned by 1audio but they didn't seem right to me. I went digging through the Mouser catalog and I'm still not sure if I found the right part. There's a lot of things I don't know. I guess sometimes solder solves the unknown just as good.
This is a bit complicated... by digikey you need to search for HOUSING's
one choice
65240-003LF FCI | 609-1263-ND | DigiKey
76347-401LF FCI | 609-4453-1-ND | DigiKey
the first is the housing, the second is the "crimp" pin (you need 2 of them). If you do not have the crimp tool, it is still possible to attach the wire, use some some other tools, and optionally solder as well.
There are plenty of SIMILAR housing and pins, or then you can also buy femal pin stripe, but there it is pretty much more complicated to attach the wires nicely.
Yes that is the only reason for those analog oscillator, less THD than any soundcard based system.