PCSGU250 function generator output is very inaccurate

Instruments PC Oscilloscopes and Recorders
1

PCSGU250 serial no. 200901458
I purchased this unit to produce fixed low frequency (0.033 Hz) triangle and square waves. I find that I cannot reliably set the output of the function generator with any accuracy. I made the following measurements which illustrate the point:

Setting         Offset    Amplitude    Output
DC+             0.00v      0.05v       0.08v
DC-             0.00v      0.05v      -0.017v
Square wave     0.00v      5.00v      -2.330v to +2.414v
Square wave     2.52v      5.00v      +0.408v to +5.147v
Square wave     5.00v      5.00v      +3.105v to +7.840v
Square wave     5.00v     10.00v      +0.863v to +10.237

The spec says the vertical resolution is 8 bits but this is not what my results show.
No adjustment of offset or amplitude was made.
The same results were obtained off load and with a load > 180kohm.

Is this unit faulty or is this standard behaviour?

2

Thank you for your comments. All this kind of user response is valuable to improve the product.
Regardless the PCSGU250 is no precision instrument it can still be improved in the future.

In next release of the PcLab2000LT software the correspondence between the displayed and actual output voltage values can be improved.

3

Clearly when the spec says ‘vertical resolution 8 bit’ then this does not equate to 0.5% accuracy.

Presumably you test these units before you ship them and there must be some criteria that they have to pass. Please advise me what these criteria are for the function generator.

When can we expect the next release of PcLab2000LT?

4

All units are tested to fulfill these specifications:
velleman.be/be/en/product/view/?id=377622

The specified “vertical resolution: 8 bits” just indicates that there are 256 different levels available between the max and min level of the offset voltage.
This “offset: from 0 to -5V or +5V max. (resolution 0.4% of full scale)” defines the ofset voltage range. Actually it is slightly wider than specified at the moment.

The updated software is ready soon - available to download today or beginning of next week.
I’ll inform you when ready.

5

I look forward to trying the new version. Thanks.

6

You can now download the updated program from Velleman downloads page:
velleman.be/be/en/download/files/

pclab2000lt_rev_1_04.zip
Version 1.04 of the PCLab2000LT software package for PCSGU250:
Release notes:
Rev 1.04: Output voltage setting accuracy improved.
Rev 1.03: Timing error between the channels fixed

7

Thanks, I will try it on Monday and report back.

8

The accuracy is improved but it does not help me that much.

I am having to set up a set of tests for my customer to come and witness in June. The tests are predefined and require certain specified waveforms. For example, one test requires a triangular waveform that swings between 0v and 5v and the next test requires a squarewave that does the same. Unfortunately when changing the waveform, the output does not just change shape but changes the levels that it swings between.

I would have expected that the output of a DAC would be fairly consistent no matter what the shape of the waveform, especially when the frequency is 16.5 mHz, i.e. one cycle per minute.

As it is, I have had to go through every test and find a value for offset and amplitude that works, and hope that these numbers will work in June. My task would be so much easier if I could set the upper and lower limits using a squarewave and know that those limits would persist if I change to a triangular wave or sinewave.

9

I’m sorry that you’ve had these problems to set up the amplitude and offset. Could you please give some measurement results to me as a reference.
I can then compare those results to my results and possibly even improve the accuracy of the voltage and offset settings of the PCSGU250.
According to my tests this generator seems to be very stable. I think you can trust the offset and amplitude settings will work in June too.

10

I have carried out a few tests using a Fluke 45 meter (calibrated 7 May 2009), which has a Min/Max function that is useful for measuring the triangular waveform.

The pulse waveform used was created in PCLab2000LT rev 1.04 and is specified as follows:

1.0(1000) -1.0(3000)The tests were carried out with the following settings:
Fine Tune settings set to their centre position.
Peak Offset and Auto Offset set to Off.
Frequency set to 16.5 mHz
Load >= 180Kohm

The test results are as follows:

Triangle Square 25% Pulse Offset Amplitude min max min max min max 5.00 10.00 0.253 10.108 0.099 9.998 0.317 10.172 2.52 5.00 0.064 5.059 0.025 5.028 0.179 5.009 0.00 5.00 -2.489 2.507 -2.528 2.476 -2.374 2.453 0.00 10.00 -4.811 5.051 -4.972 4.927 -4.668 4.889This illustrates that the min and max are distinctly different for the 3 waveforms.

I have not carried out any stability tests nor any frequency checks (I’m not sure that I have anything available to test 16.5 mHz) so I am assuming the best.

11

Thank you for these results.
It seems that the square wave is quite OK. Maximum amplitude error 1.01%.
In the triangle wave there seems to be maximum amplitude error 1.45%.
In the library waveform there seems to be some asymmetry and also maximum amplitude error is 4.43%

Triangle Square 25% Pulse Offset Ampl. Vpp Error/% Vpp Error/% Vpp Error/% 5 10 9.855 -1.45 9.899 -1.01 9.855 -1.45 2.52 5 4.995 -0.1 5.003 0.06 4.83 -3.4 0 5 4.996 -0.08 5.004 0.08 4.827 -3.46 0 10 9.862 -1.38 9.899 -1.01 9.557 -4.43

12

[quote=“wooders”]I have carried out a few tests using a Fluke 45 meter (calibrated 7 May 2009), which has a Min/Max function that is useful for measuring the triangular waveform.

The pulse waveform used was created in PCLab2000LT rev 1.04 and is specified as follows:

1.0(1000) -1.0(3000)The tests were carried out with the following settings:
Fine Tune settings set to their centre position.
Peak Offset and Auto Offset set to Off.
Frequency set to 16.5 mHz
Load >= 180Kohm

The test results are as follows:

Triangle Square 25% Pulse Offset Amplitude min max min max min max 5.00 10.00 0.253 10.108 0.099 9.998 0.317 10.172 2.52 5.00 0.064 5.059 0.025 5.028 0.179 5.009 0.00 5.00 -2.489 2.507 -2.528 2.476 -2.374 2.453 0.00 10.00 -4.811 5.051 -4.972 4.927 -4.668 4.889This illustrates that the min and max are distinctly different for the 3 waveforms.

I have not carried out any stability tests nor any frequency checks (I’m not sure that I have anything available to test 16.5 mHz) so I am assuming the best.[/quote]

The Fluke 45 is a fine instrument, however it is a multimeter not a wave analyzer, with a maximum sample rate of 20 S/s and a minimum rate of 2.5 S/s. I believe the sluggish sampling rate may be coloring your observations, even with a 16.5 mHz signal.

The square wave remains at it’s peaks for 30.3 seconds, allowing plenty of time for the meter to sample and record the minimum and maximum values. However in the instance of the triangle wave unless the meter’s sampling occurred exactly in synchronisation with the signal’s peaks it would not capture a true peak reading.

Sampling the signal over some long duration would tend to allow the relation between the sampling rate and waveform repetition rate to perhaps “match up” for one or more samples, however with a 60.6 s period this could take hours.

This does not perhaps explain the 25% duty cycle issues, however there may be less obvious similar conflicts at work. I have checked the sine, triangle, and square wave function of my PCGU1000 and found the the sine and triangle amplitudes are dead on (4.96V p-p set to 5.0V), however the square wave exhibits a bit of over and under shoot (even when properly terminated), in the order of 1.25%, fully accounting for the slightly higher p-p measurements of 5.04V or so.

13

Cliffyk, I take your point about the Fluke 45 being a multimeter, however even at 2.5S/s the error for a 10v amplitude would only be 0.067v for the triangle wave.

Having to work at such a low frequency is a real pain because it makes measurements very difficult.

My laptop has had to go to the IT dept for a few days to be “doctored” so it will be next week before I can progress this any further.

14

The link on the downloads page to pclab2000lt_rev_1_05.zip does not seem to be working.

15

OK, thanks for the quick reply.