Using an Active Bandpass Filter to Reduce Audio Generator THD
There are a number of low cost audio generators available that can provide basic test functions. Many of these units are based on the popular XR2206 function generator chip which has a bandwidth that can cover frequencies well in excess of that needed for audio work, as well as providing square and triangle wave forms.
However, most, if not all of these low cost generators are handicapped by having a total harmonic distortion (THD) that is much too high to be used for serious audio testing. So, is there a way to reduce the THD to a level of that of a lab grade generator? Fortunately, the means to reduce the THD is relatively simple. All that is needed is a stable bandpass filter with an appropiate Q that can be tuned from about 30cps to 20kcs.
Naturally, the amount of THD reduction is a function of the THD going into the filter. For example, a very high THD (say, 5.0%) going into the filter will not reduce the THD coming out to a level of a lab grade generator, however, the THD will be substantially reduced.
Based on tests using lab grade equipment, we’ve been able to reduce the THD of a XR2206 based generator as shown in the figures below. This is a reduction that puts the XR2206 (THD-wise) at the level of some lab generators and is suitable for serious audio testing.
The figures below show the dramatic THD reduction using a XR2206 function generator purchased off eBay.
|
DATE:
|
12/31/21
|
S/N:
|
UAF030
|
|
|
|
|
|
|
|
|
|
|
FREQ CPS
|
THD IN %
|
THD OUT %
|
Vin RMS
|
Vc RMS
|
Vcc
|
|
|
From XR2006
(THD out of XR2206)
|
From filter
(THD out of filter)
|
Input to filter
(Output of XR2206)
|
Output of Filter
|
|
|
30
|
1.050
|
0.082
|
0.580
|
0.901
|
15.200
|
|
65
|
1.080
|
0.076
|
0.583
|
0.904
|
15.200
|
|
125
|
1.170
|
0.092
|
0.585
|
0.904
|
15.200
|
|
250
|
1.700
|
0.078
|
0.582
|
0.914
|
15.200
|
|
500
|
1.410
|
0.073
|
0.584
|
0.914
|
15.200
|
|
1000
|
1.300
|
0.071
|
0.586
|
0.909
|
15.200
|
|
2000
|
1.280
|
0.070
|
0.587
|
0.905
|
15.200
|
|
5000
|
0.820
|
0.069
|
0.588
|
0.871
|
15.200
|
|
10000
|
0.600
|
0.051
|
0.590
|
0.892
|
15.200
|
|
15000
|
0.580
|
0.059
|
0.592
|
0.919
|
15.200
|
|
20000
|
0.570
|
0.057
|
0.592
|
0.942
|
15.200
|
So, if a bandpass filter may be used to reduce the THD of a low cost audio generator, could it also be used to reduce the THD of a lab quality generator to an even lower level?
Most lab audio generator have a THD in the range of about 0.07% to around 0.04%. How much lower can we reduce the THD from these already low levels? Let’s look at the results obtained from a Rigol function generator which already a low THD spec of less than 0.07%.
|
DATE:
|
12/31/21
|
S/N:
|
UAF050
|
|
|
|
|
|
|
|
|
|
|
FREQ CPS
|
THD IN %
|
THD OUT%
|
Vin RMS
|
Vc RMS
|
Vcc
|
|
|
From Gen:
|
From Filter:
|
|
|
15.2
|
|
30
|
.056
|
.014
|
0.500
|
1.047
|
15.2
|
|
65
|
.043
|
.011
|
0.500
|
1.043
|
15.2
|
|
125
|
.044
|
.010
|
0.500
|
1.041
|
15.2
|
|
250
|
.042
|
.0089
|
0.500
|
1.028
|
15.2
|
|
500
|
.031
|
.0064
|
0.500
|
1.018
|
15.2
|
|
1000
|
.030
|
.0054
|
0.500
|
1.014
|
15.2
|
|
2000
|
.028
|
.0052
|
0.500
|
1.009
|
15.2
|
|
5000
|
.027
|
.0028
|
0.500
|
1.004
|
15.2
|
|
10000
|
.017
|
.0031
|
0.500
|
1.023
|
15.2
|
|
15000
|
.017
|
.0034
|
0.500
|
1.039
|
15.2
|
|
20000
|
.014
|
.0034
|
0.500
|
1.082
|
15.2
|