A reduction from a theoretical 96 dB to "70-80dB" seems a lot to blame on "noise", is your test set-up noise really that bad? If so, why?
What are the "other factors"?
How have you checked that it is not injection non-ideality errors, with simulation or measurement?
Can you expand on this?
The 4195A spec. sheet claims the Dynamic R. is > 100 dB so I don't understand why your result is so poor, hence my answer that it's injection non-ideality.
But I have never used an HP 4195A personally, they are labelled "obsolete" on the company website (3 1/2" removable disks!), so no doubt there's much to learn.
They actually look quite nice.
Best wishes
David
I've never used a VNA for audio, but in my (l-band) experience all the work in getting a meaningful measurement is about calibration, using really good cal standards, knowing exactly where your reference plane is, and getting the power level right for the DUT. If you've got 160 odd dB gain, you've got to push the generator level way down, and the instrument doubtless isn't built for that.
In RF stuff I'm thrilled at getting -50-60dB S12, and -20dB S11 & S22 is seen as a very good match. That's what this sort of instrument is built to measure.
In RF stuff I'm thrilled at getting -50-60dB S12, and -20dB S11 & S22 is seen as a very good match. That's what this sort of instrument is built to measure.
Will unsubscribe now. Hiding schematics and pretending instead has nothing to do with DIY.
Best regards!
Best regards!
Because there is much more beyond the Johnson noise in the 4195A and the 41800A (1:1 probe). There's neon lights and CFLs around, polluted mains, city electrical grid, RF generators around, other EMI sources. I don't have a shielded box, so I need to set the input attenuators to -10dB to avoid channel overloading.
Please read a few pages down the 4195A specification and you will find out how Keysight specs 100dB dynamic range with 16bit A/D converters.
The most common method to increase the dynamic range is to use a dynamic input signal level and attenuators. It's like the dynamic range window slides around the measured value. Another one is using wideband precision logarithmic amplifiers as input stages. I have none of these. One day I'll try building my own logarithmic probe using a LOG114 IC from TI or other similar. Absolute precision is not a big issue, since the measured value is always a ratio.
The same limitation in dynamic range applies for measuring an opamp open loop gain and open loop unity gain frequency (no injection issues there). The dynamic range of the 4195A + 41800A + unshielded DUT is about the same.
Enough said.
Nor I, but I have considered it for some specific tests I would like to do.
But I would have to borrow or even buy a VNA, do you still have access to one and the time?
I would love to learn if it is worthwhile, please post if you ever try it.
Yes, of course, but "70-80dB", as you wrote, is poor S/N even for a basic DIY audio set-up.
Is your lab really worse than even a DIY facility?
The 4195A should be able to do some kind of noise spectrum, if you post the baseline then we have a basis for discussion.
Best wishes
David
Last edited:
I've got an Anritsu MS2036C in the lab at work, but it only goes down to 5KHz, and I imagine it'd be really revolting down there. I remember working my way through all the synths and spec-ans in the lab years ago - every last one was lucky to do 80dB DNR and I don't recall we had anything that'd go as far as 1KHz. I ended up borrowing an antique aa501a for measuring THD from another lab.
The usual use I have for VNAs is characterising things in 50 ohm systems. On the odd occasion I've measured things at other impedances, I've gone to rather a lot of trouble to build an impedance transformer to match it to the 50 ohm VNA impedance. I'm not certain how valid results would be without some degree of impedance matching - I think at the least you'd need to buffer things to get 50 ohms, and once you've done that the only measurement that makes sense is S21, so you might as well simply use a CRO or ADC.
My HP3585B makes a nice scalar analyser using the tracking generator, so that's kept me happy. Here's a pretty typical plot of closed-loop gain of one of my amps out to 10MHz. In this plot I do ~75odd dB dynamic range, but I could push that down to ~90dB without trouble.
Last edited:
The HP 4195a specs say down to 10 Hz with 0.001 Hz resolution, so, better for audio work.
HP even explicitly mention analysis of 50/60 Hz power line side-bands - kind of over-kill for that😉 but indicative of low frequency capability. Probably part of the reason why they are still decent money, despite their years.
So I expected better noise performance in the low frequency band, 70-80 dB seems fairly mediocre.
Obviously different at RF, as you say.
Best wishes
David
Last edited:
- Status
- Not open for further replies.