Rtb2004, what am i doing, or understand, wrong?

[Richard576]

Perhaps to much info but i want to be complete as possible not to leaf out any info that might be relevant.

I am using my rtb2004 as a glorified Vpp meter @455 khz to measure a problem i got with a lowe receiver in the ams detector.
Checking the voltages needs a frequency depended volt meter which i dont have so i want to use the FFT in stead so i can see and measure the voltage level only at 455 khz.
I got it all worked out,the math function transforms the Veff (RMS) via multiplication over (2xsqrt2) from the Vrms to Vpp as the lowe tech manual gives all values in Vpp and not Vrms.
Thats all dandy and works.
Now what i see is there is a difference between sample mode and high res mode.
When i feed it 455 kHz 20 mVpp sine from the generator i measure a near perfect 19.85mVpp in the FFT in sample mode but if i switch it to High res mode it drops to 15.83 mVpp.

(Edit, i just checked:)
If i measure the C1 channel directly, so not through the math function i see the same image, high res gives a 1/4 lower result compared to measuring in sample mode.
So its not the math function bothering.

Why does the high res mode give a false result while i expect a more accurate result because of the increased bitcount?

Latest firmware, auto rbw on or off makes no difference.
Manual
Center freq 455 kHz
Span 85 kHz
Rbw 70.9 Hz
W 18.35 ms

Auto
Rbw 425 Hz
W 3.06 ms

I tried hanning, hamming,blackman,flat top,rectangle, all no difference the measured voltage in high res remains 1/4 low.

Whats the problem here?

Thanks Richard

[Richard576]

I took a little time and effort to get what you are saying but after some testing i understand now, the drawback of a small span is the lack of data and therefore the measurement goes wrong.
But as you can not choose the timebase in the fft of the rtb the only thing i can do is use a wider span and thereby lowering the timebase, it means i would have to increase span to about 4 MHz to get timebase below 1 ms.
I raises the question if the solution to the problem is not worse then the original problem circumvention by just not using high res mode?
As i see it now the options are high res mode on a 4 MHz span or sample mode on a 10.6 kHz span, i am inclined to think sample mode on a small span would be best also given the idea that at 4 MHz span there could be other signals hidden in the coarse resolution which i could miss and not notice.
I will do some more testing to see if i can verify one or the other.
For now at least i can get back to my original problem and trying to repair my receiver, i think sample mode will be sufficient accurate for these measurements.

Thanks for your help and insight 👍🏻

(Edit) i noticed to late you posted a screen shot, the thing is: i cannot make the settings you did, maybe because of a firmware difference, the latest firmware has a totally different way of handling memory and timebase in the fft, it makes that the fft is more usable at lower frequencies but probably at the expense of handling overall.
If i try to make you settings center 455, span 572 , rbw 320 i get a window of 4.5 ms and a time base 5.885 ms
If i change the window to 12 ms like yours my rbw goes to 120 Hz but timebase stays the same at 5.885 ms.

I can not get the timebase down to 1 ms by adjusting rbw or window, i have to increase span a large amount.
I will make some screenshots...

[Richard576]

Some screenshots of my efforts to replicate your settings, i cant.
The new firmware makes that changes to rbw or windows does not change timebase, the only way now to change timebase is through span.
With that there is now an "improved" auto rbw setting on the fft settings panel, if you choose "automatic rbw" you are rewarded with a 4x smaller timebase on the very low frequencies(audio and such), it is now practically possible to look at 50/60 Hz without a timebase of 40 seconds resulting in a single scan time of 12x40 seconds=480 seconds.
With the new auto setting it does a 8 second scan on 50 Hz which is (just) bearable.

I think this is all quite miserable just to make life easier for people who cant handle the caveats of manual adjusting the settings.
Before i had a HMO1202 which stills has al these manual settings like record length,window,,timebase so you have to adjust all these manually, it is more work but gives superior results, makes it possible to measure from 50 Hz up to max with maximum results.
This latest update also make the scope more buggy in the fft, i find myself now regularly doing a factory default reset because it is somehow stuck on a setting.

 Also came to the conclusion that the rtb which i bought and thought to be superior to the HMO1202 is in fact inferior on a lot of things to the HMO, not having a 50 ohm on a 300 mMHz scope is just one of these things, what were they thinking when they made this thing?
Now with the latest firmware it is buggy where the HMO i never found any bugs at all.
But thats just me ranting 😅

Your thoughts on the latest firmware?, i assume you dont have it?

[strawberry]

10bit ADC with up to 16-bit with high resolution decimation
higher resolution = lower bandwidth
DC Y full scale accuracy +-2%
FFT accuracy not specified

10bit ADC 5mV/div = 50uV/bit resolution ( 50mV signal 0.1% error )( 20mV signal 0.25% error )

[Richard576]

I got some funny stuff going on, i did exactly as you described, first did a factory default, adjusted gen and scope to what you say additionally like to mention i have record length set to auto, i guess you also have it set to auto.
 When i get to the fft it still looks like yours but then the scope wont let me make adjustment to the values you do, no matter from which setting i approach it i can not get at a center of 455 your span, rbw or window, it will not let me as long the timebase is sort off "locked" on the 1 ms setting.
If i then try to change record length i can change from auto to a specific value, it doesnt matter what value i choose, as long as it is not on auto the timebase then jumps to 5.33 ms
I then can change the setting to yours but then timebase is at 5.33 ms so i have the inaccuracy on highres.
The only way to avoid that is then increasing span to 4 MHz, the timebase is now not locked anymore at 1 ms and moves when i change start,stop,span,rbw or window.
Not also that after changing the timebase i can only (almost) get to your values provided i choose flattop as fft window, with the other fft windows i cant get to your values.
I made several screenshots showing the process.
I am getting inclined to re do the firmware update, it seems there is something wrong?...

Photo 1-5: trying to make your settings but scope wont let me no matter what value i try to change to your values.
Photo 6 in scope mode record length is set to auto giving a rate of 10.4 MSa/s
Photo 7 in fft mode the record length is still adjustable but at auto giving a rate of 10.4 MSa/s
Photo 8 i change record length to (no matter what) 200 Ksa and timebase changes to 5.33 ms.
Photo 9 the record length adjustment is now greyed out,cannot be changed anymore and is fixed at 131 Ksa.
Photo 10 now i can make your settings but the timebase is set at 5.33 ms and highres is inaccurate

[Richard576]

O wow, can you believe it?
I re did the firmware update and now everything is honky dory, i can make the settings you suggested 🙄
I wonder i any of the other strange behavior i encountered is also fixed?

After the firmware update my scope really seemed much worse then it was, so much hanging values that could not be changed, i had to do regularly a factory default reset to get it going again 🤬
And now i discover probably the firmware update didnt go well....
This explains a lot 😂

Well thank you👍🏻, your examples pictures did a lot for me narrowijng down i had a firmware problem, thank you so much.
Now i can focus on fixing my radio ☺️

[Richard576]

10bit ADC with up to 16-bit with high resolution decimation
higher resolution = lower bandwidth
DC Y full scale accuracy +-2%
FFT accuracy not specified

10bit ADC 5mV/div = 50uV/bit resolution ( 50mV signal 0.1% error )( 20mV signal 0.25% error )

I am trying to picture what you are telling me but have a hard time with the math, one thing i do know is that it is uncertain what high res mode does, there is no visual feedback on what bitrate it works besides it is "up to 16 bit",  so it is unclear what the advantage is besides seeing the noise floor drop in the fft or the waveform cleaning up, or sometimes not concluding that in such a case apparently the bitrate does not improve a lot, the performance of high res is variable depending on how much data there is to spare from what I understand how it works.
The other thing i know is that a scope is originally (an analog scope) not really meant to be used as a quantitive measuring tool but is a quality measuring tool, it is meant to look at waveforms and such and not to measure nano volts, but with digital scopes that has changed and with the implementation of fft in scopes and well designed and accurate front end it has become more and more a glorified volt meter.
It is just that things like the high res mode make it (for me) uncertain what i can expect in accuracy.