2204A accuracy / calibration?

Post any questions you may have about our current range of oscilloscopes
Post Reply
AC/DC
User
User
Posts: 2
Joined: Mon Feb 20, 2017 8:27 pm

2204A accuracy / calibration?

Post by AC/DC »

I have measured my 2204A against a voltage reference and it reads a bit high on:
Channel A in the voltage range 5,10,20.
Channel B in the voltage range 10,20.

I wouldn't mind the guarantee, so if I'd open the device what would I find? More specific, is it possible to calibrate the ranges independently, or is there one pot per channel affecting all ranges?

Gerry
PICO STAFF
PICO STAFF
Posts: 1145
Joined: Mon Aug 11, 2014 11:14 am

Re: 2204A accuracy / calibration?

Post by Gerry »

Hi AC/DC,

If you're not applying any offset in PicoScope 6, and the error is more than 103% of your voltage reference then it's classified as faulty, and you should be able to return it to Pico for repair. If it's less than 103% then it's functioning according to it's spec (as it hasn't been designed for greater accuracy than that).

Unfortunately, applying a correction is not as simple as tweaking a pot. However, if you want to trim out as much of the error as you can, without adding more error elsewhere, you would need to apply more than one reference voltage value to gauge the linearity across the range. The reference voltage would have to be, as a bare minimum 3 (preferably 10) times more accurate and have been calibrated, recently enough, over the values that you would be applying to check the linearity (so a switchable reference, rather than a single one). If the error is linear enough then you could theoretically create a custom probe file for that range in order to apply a correction using scaling. However, if after having established that there is room to improve the accuracy, it would be better to send the PicoScope back to us to make the adjustment, as the granularity available to our test and repair team is finer than a scaling file (and you don't invalidate the warranty). If you're not able to establish the linearity as well as the single point error, then it's probably not going to be effective to apply a correction, and not cost-effective (based upon the cost of the 2204A) to send it to us in order to establish if one is necessary.

It's always good practice in Test and Measurement to do whatever you can to increase your confidence level in the uncertainty and error of the measurement, (especially as you, the user of the equipment, are in the best position to do that on a regular basis) and doing so, using something such as a calibrated voltage reference will certainly help with that. However, it's also always worth bearing in mind that finding the error yourself is a whole different prospect to correcting it yourself.

Regards,

Gerry
Gerry
Technical Specialist

AC/DC
User
User
Posts: 2
Joined: Mon Feb 20, 2017 8:27 pm

Re: 2204A accuracy / calibration?

Post by AC/DC »

Thank you very much, Gerry.

Yes, using a calibration service is in the hobby segment not really an option, so I've bought one of those very cheap voltage refs, switchable between 2,5 / 5,0 / 7,5 / 10V. I know you Pro's will laugh at me, and I've learned that the ref I'm using has fake documentation about the stated values. Though, in this low budget segment it's the only option to get more accuracy. I've ordered two others, where the user reviews state that the values seem to be measured for real. With 3 refs I will see a bit if their stated values make sense. I know, at the end I can't count on it. It is more a "best accuracy I can get for really low budget"-approach. I swear I will not repair a nuclear plant using my 2204A. ;-D

Thanks for the custom probe tip, very helpful! This is for my needs a very good solution as I don't have to recalculate the values after measuring. :-) And I guess it's more than enough for my needs, compared with a real calibration, which would, if I understood everything correctly, not improve the readings in all ranges.
My current v-ref, my UT61E (also cheap, but known for good accuracy), and my 2204A have all the same reading at the stated 2.49924V of the ref: 2.499V. Only channel A is a few counts higher. My probe table now looks so, at 22.5°C, always adjusted with the lowest possible measurement range:

Channel A:
0 = 0
2504 = 2499 (+0.2%)
5034 = 5000 (+0.7%)
7568 = 7500 (+0.9%)
10100 = 10001 (+1%)

Channel B:
0 = 0
2499 = 2499 (+0%)
5011 = 5000 (+0.2%)
7521 = 7500 (+0.3%)
10020 = 10001 (+0.2%)

If the voltage ref and the UT61E are not lying too much, and I don't think so, then the base accuracy of my 2204A is already quite impressive, especially for channel B, wow. And with the customizations I guess I'm now always well under 0.5% error, nice.

I'm completely "blind" below 2,5V because that's the ref's smallest voltage, and I didn't found any refs below 2,5V. In case the Pico is calibrated against 2,5V (it's spot on there) then I guess the readings below 2,5V might me too low? More specific, I'm *usually* measuring only from 1V to 10V. Can I expect that the error from 1.01 - 2,5V ist not higher than it is at 2,5V?

The Pico is switching relais at 1V and below. Does it have to do with the measurement range? Is it possible that the error is greater in that range because it's probably(?) a "complete different thing"?
For my needs it would be nice to know/guess what my 2204A's error is at around 300 mV. Can I make any assumption based on the results in the 2,5 - 10V range?


As a side note I also want to thank Picotech for offering affordable entry hardware at that quality. I've checked other vendors, and I think the 2204A is the best at that price point. If I have to moan something it is the Mac software, which exists for years, but not really moving on. I have to use a VM, with it's own problems. :-(
Though, your strategy seem to work. I wouldn't have bought an oscilloscope at all if they where more expensive. But now that I've learned a bit with the 2204A I already want 4 channels, damn. :-) Unfortunately the cheapest is 4x the price of the 2204A. Actually I would just buy a second 2204A, available for under 100 (without probes), cool. But the software cannot handle 2 devices, so I'm a bit stuck. Looks like I have to sell grandma. Anyone interested?

Gerry
PICO STAFF
PICO STAFF
Posts: 1145
Joined: Mon Aug 11, 2014 11:14 am

Re: 2204A accuracy / calibration?

Post by Gerry »

Hi AC/DC,

To answer your questions:

Factory calibration would improve the accuracy in all ranges, only if all ranges are out by more than +/-3% +/-200uV.

Unfortunately, you can't rely on any expectations or assumptions, as non-linearity applies to any range or value, so the error at 1V could easily be higher (in % of full scale) than the error at 2.5V.

The relay is switching the gain stage for input range, and the gain stage can also contribute to the error.

Let's break down this issue of self-calibration for entry level PicoScopes:

What you need to bear in mind is that you should be looking to increase your accuracy of measurement without reducing the confidence level in the accuracy of the measurement. If you are using uncalibrated references from questionable vendors, are you really doing that?

I can appreciate that you want to do this on a budget, but the cold hard fact is that sometimes satisfying a requirement may just not make economic sense. If you are looking to establish the linearity using a switchable voltage source, then that needs the following;
  • Coverage of the complete range of measurement that you'll be using (so it needs to be something like a Voltmeter, if you want to use the lower ranges as well).
    An accuracy of less than 1%, and a low drift.
    A recent enough calibration.
However, if you're calibrating a Voltmeter on a regular basis then it defeats the object of avoiding calibrating the PicoScope.

If, on the other hand, you don't need the full range of inputs and you're using a reduced range, switchable, low drift voltage source from a reputable vendor, then you should only need to calibrate the voltage source very occasionally. The reputable vendor boosts your confidence in the drift specification, so that, along with how frequently you use it, the usage cases and environments, you have some idea of when to re-calibrate. Another reason for using a reputable vendor is that you can have the confidence in being able to factor in any stated specifications that contribute to the error and uncertainty of the voltage source (e.g. any hysteresis that may be significant). Also, while the IC used for the voltage reference may be high precision, a reputable vendor is less likely to use a low cost, poor power supply, and/or a poor board design/layout that contribute to errors.
So, under these circumstances, and only these circumstances, with the voltage source having been calibrated (where necessary), if you are able to verify that the non-linearity is small enough, and accurate enough, then you could apply a scaling correction with custom probes, to perform a course calibration/correction.

Alternatively, if the voltage source is reduced range, but you need to measure outside of the reduced range, or if the voltage source is single point, and you have high confidence in it's accuracy, then you can use it to establish if the PicoScope is faulty (beyond the +/-3% +/-200uV) to send it back for repair, knowing that you won't be making any measurements with questionable accuracy when the PicoScope is returned.

Also, to expand on your last point, PicoScope 6 will not connect to 2 PicoScopes simultaneously for good reasons, including the fact that you wouldn't be able to synchronise the capture of waveforms between the 2 hardware scopes. However, I'm glad that you're happy with the 2204A, and that it has brought you affordable data acquisition. We think that, when coupled with our PicoScope 6 software and the Software Development Kit, it is a great product for it's price point. However, I would be really careful about placing too much significance on the accuracy of your readings without justified confidence in the accuracy.

Regards,

Gerry
Gerry
Technical Specialist

Post Reply