Here is a picture of a zoom-in on a 1 kHz 20Vpp sine wave. The scope scale is 10 mV/100nsec per division.
FY6800 on a sine wave is about the same result 12bit
On the ramp signal, the resolution is 14bit
In the first case, the sine pitch is 4mV, in the case of 1mV ramp
That's very interesting. It looks like FY6800 uses 12 bit NCO to generate sine wave.
Can you check it on different sine wave frequencies (with different rise time), for example 100 Hz and 100 kHz?
That's very interesting. It looks like FY6800 uses 12 bit NCO to generate sine wave.
Can you check it on different sine wave frequencies (with different rise time), for example 100 Hz and 100 kHz?
think the DAC is 14 bits, but the bit depth limits software.
Yes, I measured the value of the step at different frequencies of the generated sine 10 kHz, 1 kHz, 100 Hz, it does not change
So the real resolution for sine wave is just 12 bit
14 bit NCO requires more FPGA resources, and probably doesn't fit into EP4CE6, so they decided to limit resolution with 12 bit instead of use more expensive EP4CE10 or EP4CE22
So the real resolution for sine wave is just 12 bit
The Chinese are cunning, it’s good that they didn’t slip 10 bits.
It's a good idea to test an up-ramp. For 8192 horizontal points one gets 8192 vertical points. So the vertical resolution should be 13 bits in this case. Here is the picture for an up ramp on FY6600. The scope resolution here is 5 mV/100 nsec per division. The steps are indeed about 2.4 mV=20V/8192. So the DAC is not at fault. It is the number of points in the waveform that is creating the steps.
This is not purely FeelTech limitation. Rigol and Keysight are also limited to some (although lesser) extend by the waveform length.
https://www.eevblog.com/forum/testgear/new-rigol-16-bit-function-generators-dg800900-series/msg2452248/#msg2452248
The steps are indeed about 2.4 mV=20V/8192. So the DAC is not at fault.
10 V / 2.4 mV = 4166, it's very close to 4096. So, it seems like 12 bits. With 12 bit you will have 2.441 mV step.
Do you measure it with 50 Ohm terminator on the oscilloscope connector? That's very important.
It's a good idea to test an up-ramp. For 8192 horizontal points one gets 8192 vertical points. So the vertical resolution should be 13 bits in this case.
Yes, I was mistaken when I switched the waveform to ramp, the amplitude became 10V instead of 20V. Thus, it is not 14, but 13 bits, which are limited by the amount of memory.
No, there is no 50 Ohm terminator and it is not needed. It would simply attenuate both the signal and the steps by a factor of 2. The scope is calibrated for Hi Z load, so 20V pp is indeed 20 Vpp without terminator.
No, there is no 50 Ohm terminator and it is not needed. It would simply attenuate both the signal and the steps by a factor of 2. The scope is calibrated for Hi Z load, so 20V pp is indeed 20 Vpp without terminator.
such measurement is incorrect. With Hi-Z your oscilloscope will show you random amplitude, you cannot believe it.
100 ns step uses about 10 MHz bandwidth, at >10 MHz you can get very high amplitude mistake on 1 meter cable with HiZ input.
With Hi-Z your oscilloscope will show you random amplitude, you cannot believe it.
Do not invent, everything is correct, it shows without a terminator.
With Hi-Z your oscilloscope Do not invent, everything is correct, it shows without a terminator.
No. With no terminator, there is 1M +15pF impedance on the cable end.
It means that your cable works as impedance transformer with random transformation ratio. In simple words, your 1 meter cable works as step-up or step-down transformer with random and unknown transformation ratio.
Since we-re talking about 100 ns steps, you're needs to take into account 10 MHz bandwidth. But since each step has square waveform, actual signal bandwidth is much more higher than 10 MHz. It's about 100-1000 MHz.
So, you can assume unterminated measurement just as random value... You cannot believe amplitude measurement at the end of transmission line with no proper termination.
No. With no terminator, there is 1M +15pF impedance on the cable end.
Since we-re talking about 100 ns steps, you're needs to use 10 MHz frequency.
I can set a time step longer if I use a low frequency and it will not change anything.
In addition, the inclusion of a 20 MHz bandwidth limit also does not spoil the signal, just steps with a longer front, but they can also be seen with the same vertical step.
pantelei4, since there is no proper impedance match, this is measurement at the output of transformer with random transformation ratio and random frequency response. It's transformation ratio will depends on many factors, include cable length and frequency.
All what you know, is that all amplitudes on your oscilloscope will be scaled with random factor which also depends on the frequency. So, there is no way to measure amplitude in correct way with no proper termination at the end of transmission line.
since there is no proper impedance match, this is measurement at the output of transformer with random transformation ratio and random frequency response. It's transformation ratio will depends on many factors, include cable length and frequency.
The level of quantization will not change.
Can show you the vertical step of a linearly increasing signal with a frequency of 1 Hz?
Can show you the vertical step of a linearly increasing signal with a frequency of 1 Hz?
I suggest to avoid 50-60 Hz and 0...100 Hz range, because some oscilloscopes have amplitude distortion at such frequency.
I think 1000 Hz will be good enough. So, the time interval between steps should be about 1 millisecond.
But anyway, proper 50 Ohm termination is a better and much easier way to get proper amplitude measurement.
I suggest to avoid 50-60 Hz and 0...100 Hz range, because some oscilloscopes have amplitude distortion at such frequency.
I have a Rigol and it works correctly, and not like your Siglent.
I have a Rigol and it works correctly, and not like your Siglent.
yeah, but who knows, may be Rigol has another issue...
may be Rigol has another issue...
Of course there is nothing perfect.
I added some pictures on how to connect signal generator to measure amplitude properly.
This is strongly required if you want to know real amplitude value instead of random. Don't confuse, this is not just a decrease in amplitude by half. This is the only way to know real amplitude with no random scale factor.
For whom?
you already know that. But many peoples trying to use direct connection of signal generator to the oscilloscope with no impedance match and have measurement issues because of that. And they don't know how to connect signal generator for proper amplitude measurements So I think it will be useful information for these peoples
Even people who knows that, often forgot it and make measurement mistake
Thus, the JDS6600 with a memory length of 2048 points has only 10 bits of resolution on the sine and 11 maximum on the ramp. Unfortunately, I can’t verify this, because I broke it against the wall.
For whom?
you already know that. But many peoples trying to use direct connection of signal generator to the oscilloscope with no impedance match and have measurement issues because of that. And they don't know how to connect signal generator for proper amplitude measurements So I think it will be useful information for these peoples
Even people who knows that, often forgot it and make measurement mistake
This is not impedance "mach" , just loading the signal at the scope input with 50ohm or whatever lowish ohm value will eliminate some of the the probes impedance effect . But of course you can't use 10x attenuation , the signal is too low .
scope input with 50ohm or whatever lowish ohm value will eliminate
No, the only way to get correct amplitude is to use exactly the same input impedance as your cable impedance. If you use 50 Ohm coax cable, you're needs to use 50 Ohm terminator. With any other value, for example 30 Ohm or 70 Ohm, amplitude will depends on frequency and cable length, so you cannot believe it.