Precision Bench

Mike Allen at the bench Photo by Carl Liss.
Precision Bench
Michael Allen came up with a series of exacting precision tests using his HP signal generator and balanced input oscilloscope. He had time only to test eight of the 32 1/3-octave EQs plus a vintage biamp EQ210 stereo 10-band he brought as a personal reference. A stowaway! He examined noise on the outputs both loaded and unloaded, line driving abilities, CMRR properties and silk-screening accuracy.

Let's go straight to the results!
Audient ASP231 - best line driver; best overall measurements
Rane DEQ 60L - best resistance to fader interaction
BSS FCS-960 - great CMRR specs; best square wave symmetry
TDM 30GE-2 - great CMRR specs; no peak lights and no fader detents
ART HQ-31 - poor silk-screen level accuracy; softest waveform at clipping
Rane ME 30 [early unbalanced version] - most noise at idle
biamp EQ210 [vintage] - crazy hot unloaded output voltage at nominal 'unity' setting
Behringer GEQ3102 - least noise at idle; poor CMRR performance
Nady GEQ231 - worst line driver & CMRR; worst overall measurements

Here are Mike's Excel spreadsheet results for downloading.
Left click to view, right click to 'Save As'...

Testing Parameters - EQ shootout - 12/17/03

By: Michael Allen

Test Objectives:
- To provide platform and parameters for testing of candidate Graphic Equalizers. To as closely and carefully as possible, mirror or simulate real usage conditions found in user environments.
- To perform tests that cannot be measured by SMAART, the interfacing sound module, or other interface, application, signal, and patching limitations generated by the platform used for SMAART.

Specific Objectives include:
- Candidate EQ residual noise measurements with and without output loading;
- Accuracy of unity gain signal output of Candidate EQ;
- Amplitude and accuracy of Candidate EQ output prior to clipping when driven by a signal - unloaded, loaded, and loaded with long cable run;
- Candidate EQ Fader Boost and Cut control marking accuracy;
- Test for Maximum output to Input ratio prior to clipping with output load on Candidate EQ at three frequencies;
- Check for Candidate EQ adjacent fader interaction with given signal at full boost - prior to clipping;
- Check for CMRR effectiveness of Candidate EQ at both Power line frequency and midrange frequency;
- Check for accuracy of midrange square wave across all faders.

Test Scenarios:
To measure performance in the following categories, with the elaborated test configurations for each:

1.) Noise output (measured in Peak-to-Peak mode): with 5’ balanced cable patched to input, with 100 Ohm input load, no signal. Frequency faders on flat, gain set to unity, and 5’ balanced output cable to differential Oscilloscope input. Record Results.

2.) Noise output with 600 Ohm load on output (measured in P/P): with 5’ balanced cable patched to input, with 100 Ohm input load, no signal. Frequency faders on flat, gain set to unity, and 5’ balanced output cable connected to differential Oscilloscope input. Record Results.

3.) Unity Gain [Accuracy] (measured in P/P): with Signal generator calibrated directly with Oscilloscope to a 1KHz sine wave set to 2VAC P/P. Patch back into 5’ input cable, and with all Frequency faders on flat, gain set to unity on Candidate EQ, check output at end of 5’ balanced output cable connected to differential Oscilloscope input. Record Results. Ideal Candidate EQ shall be within 0.1V P/P of input value.

4.) Maximum Unity output (measured in P/P): with Signal generator calibrated directly with Oscilloscope to a 1KHz sine wave set to maximum output prior to Candidate EQ unit clipping. Patch back into 5’ balanced input cable, and with all Frequency faders on flat, gain set to unity, check output at end of 5’ balanced output cable connected to differential Oscilloscope input. Record Results.

5.) Maximum Unity output w/ 600 ohm load (measured in P/P): with Signal generator calibrated directly with Oscilloscope to a 1KHz sine wave set to maximum output prior to Candidate EQ unit clipping. Signal fed into 5’ balanced input cable, and with all Frequency faders on flat, gain set to unity; check output at end of 5’ balanced output cable loaded to 600 ohms connected to differential Oscilloscope input. Record Results. Ideal Candidate EQ shall show no more than 10% deviation on loaded vs. unloaded value.

6.) Maximum Unity output w/ 600 ohm load and 300’ output cable: (measured in P/P): with Signal generator calibrated directly with Oscilloscope to a 1KHz sine wave set to maximum output prior to Candidate EQ unit clipping. Signal fed into 5’ balanced input cable, and with all Frequency faders on flat, gain set to unity, 5’ balanced input cable; check output at end of 300’ balanced output cable loaded to 600 ohms connected to differential Oscilloscope input. Record Results.

7.) Fader Boost silkscreen accuracy: Signal generator set to a 1KHz sine wave set to a mid-level output amplitude (not at extremes), patched to 5’ balanced input cable, gain set to unity, and Candidate EQ 1KHz fader boosted to first index marking indicated by unit silkscreen. Input signal to Candidate EQ measured by voltmeter, and output of Candidate EQ measured by voltmeter. Record Results, convert to db value. Compare to index value as set by Candidate EQ. Ideal Candidate EQ shall match Silkscreen to measurement within 10%.

8.) Fader Cut silkscreen accuracy: Cut the Candidate EQ 1KHz fader to the first indicated silkscreen marking. All other factors and actions mentioned remain the same.

9.) Maximum Fader output at 100, 1KHz, and 10KHz w 600 ohm loading: Boost Candidate EQ fader at each of the indicated frequencies to maximum value, and set Signal generator to same frequency as fader boosted, and to maximum output prior to Candidate EQ unit clipping as indicated on Oscilloscope. Signal fed into 5’ balanced input cable, gain set to unity; check output at end of 5’ balanced output cable loaded to 600 ohms connected to differential Oscilloscope input. Input signal to Candidate EQ measured by voltmeter, and output of Candidate EQ measured by voltmeter. Record Results. Convert to db. Repeat for each frequency. Ideal Candidate EQ shall exhibit maximum boost within 5%.of stated value.

10.) Maximum Fader output at 100, 1KHz, and 10KHz w 600 ohm loading and 300’ cable: Same as above test, with 300’ Balanced cable in series with output, terminating in 600 Ohm load.

11.) 2nd adjacent fader test: With 1 KHz boosted to maximum, and signal generator to maximum output prior to Candidate EQ unit clipping as indicated on Oscilloscope, perform boost and cut on the second fader away from the 1 KHz fader on each side. Observe the amount of change in waveform. Moderate to large changes in waveform (>30%) are considered grounds for failing. Small to moderate (<10%) changes in waveform are considered passing. Record Results.

12.) Input CMRR, 60Hz, 5V P/P test: Signal Generator set to 60 Hz, 5V P/P. Input to Candidate EQ is the positive component of the balanced signal fed to both + and - inputs. Balanced output of Candidate EQ is then fed into the balanced Oscilloscope input, and measured for P/P signal. The ideal Candidate EQ shall have little to no output - other than typical residual noise. Record Results.

13.) Input CMRR, 1KHz, 5V P/P test: Signal Generator set to 1 KHz, 5V P/P. Input to Candidate EQ is the positive component of the balanced signal fed to both + and - inputs. Balanced output of Candidate EQ is then fed into the balanced Oscilloscope input, and measured for P/P signal. The ideal Candidate EQ shall have little to no output - other than typical residual noise. Record Results.

14.) Square Wave Linearity test: Feed a 1KHz 0.5V P/P Square wave signal into a Candidate EQ, and observe changes to signal as different frequencies on Candidate EQ are boosted and cut. Frequencies above 1KHz should show Linear changes in overshoot of waveform, size and shape of ripple, and frequencies below should show linear changes to 1KHz Waveform in the form of tilting to waveform. This test is probably the most subjective, and difficult to quantify, but in general, a Candidate EQ which passes all the above tests generally fares well in this test, whereas a poor performer will generally not fare well in this test. It should be noted as well that this test was added at the request of the EQ Shootout committee as a last minute impromptu item.

General Notes:
1.) General comments about Equipment used and setup used available on test spreadsheet.
2.) All measurements done on Oscilloscope are to within the observable limits of the vertical index grids and scale used. Therefore, large signal measurements done on Oscilloscope will have greater tolerance margins than those done at low signal scale.
3.) Voltmeter used on all measurements where RATE OF CHANGE is the objective, owing to the DVM’s greater observable resolution. All measurements converted from Voltage numbers to db numbers, using the following formula:

                V Large
20 Log       ----------- = ___ db
       10       V Small

	Biamp EQ210	BSS FCS-960	Audient ASP231	Nady GEQ231	Behringer GEQ3102	TDM 30GE-2	ART HQ-31	Rane ME 30	Rane DEQ 60L
Noise output, no load	2.2mv	0.3mv	0.3mv	2.0mv	0.2mv	0.3mv	0.6mv	6.0mv	4.0mv
Noise output, 600 ohm load	2.0mv	0.2mv	0.29mv	0.65mv	0.19mv	0.29mv	0.4mv	5.0mv	3.5mv
Unity accuracy at 2v input	4.2v	2.0v	2.0v	3.6v	1.95v	2.4v	1.5v	2.0v	1.9v
Max output, unity, no load	50v (!)	22.5v	28.5v	22v	23.5v	26v	19v	24v	24v
Max output, unity, 600 ohm load	23v	20v	26.5v	7.8v	20v	21v	12.75v	17v	18.5v
Max output, unity, 300' cable, 600 ohm load	21v	20v	26.5v	7.8v	20.25v	21v	12.75v	17v	18.5v
Fader boost silk-screen accuracy	+3=+2	+3=+1.5	+2=+2	+6=+9.5	+4=+4.5	+4=+6	+2=+7	***	+2=+0.5
Fader cut silk-screen accuracy	-3=-1	-3=-1	-2=-3.6	-6=-9	-4=-6	-4=-7	-2=-4.75	***	-2=-7.6
Max fader boost, 100Hz	+15.5dB (125Hz)	+14.7dB	+16.9dB	+11.9dB	+13.7dB	+14.5dB	+14dB	+11.8dB	+12.6dB
Max fader boost, 100Hz, 300' cable	+15.3dB (125Hz)	+13.8dB	+16.9dB	+11.9dB	+13.7dB	+14.5dB	+13.9dB	+11.8dB	+12.6dB
Max fader boost, 1kHz	+16.2dB	+13.4dB	+16.8dB	+12.2dB	+13.6dB	+14.4dB	+14.1dB	+11.7dB	+13.1dB
Max fader boost, 1kHz, 300' cable	+15.3dB	+13.4dB	+16.8dB	+12.1dB	+13.6dB	+14.4dB	+12.6dB	+11.7dB	+13.1dB
Max fader boost, 10kHz	+7.4dB (8kHz)	+15.8dB	+18.7dB	+14.1dB	+16.0dB	+16.1dB	+12.8dB	+13.9dB	not tested
Max fader boost, 10kHz, 300' cable	+7.4dB (8kHz)	+15.8dB	+18.7dB	+13.8dB	+16.0dB	+16.1dB	+12.6dB	+13.8dB	not tested
Second adjacent fader	poor	marginal	okay	marginal	marginal	okay	marginal	okay	Wow! best
5v CMRR @ 60Hz	2.0mv	0.3mv	0.7mv	10mv	1.1mv	0.5mv	1mv	0.7mv	not tested
5v CMRR @ 1kHz	3mv	1mv	0.9mv	150mv	23mv	1.5mv	2.1mv	2.5mv	not tested
Square Wave Linearity	fail	pass	marginal	fail	fail	marginal	marginal /pass	marginal /pass	not tested

If you'd like to take a look at Mike's bench test routing switch in AutoCAD format you can check it out here.

Test bench gear. Photo by Michael Allen.