QuantAsylum QA400 and QA401

Member
Joined 2004
Paid Member
QA400 plus interface

I have started making the cables to test the QA400 with the interface board. Making proper shielded BNC to minijack cables (cheap stereo to cheesy bnc's) really got the noise down. Here are a few measurements of the QA400 plus the interface in loopback.

The first is in loopback with the gain switches at unity (+6 dB for the output because of the X2 nature of the device) The external differential voltage is approx 1.2V

The second with both at X10 and the drive from the QA400 reduced. The external voltage is around 7V

The third is as the second but one side of the diff output is grounded.

The driver and the receiver get in trouble when the voltage goes above 7V RMS. I will rescale things around that limit.
 

Attachments

  • QA400 + QA interface  loopback at 1.2V ext .png
    QA400 + QA interface loopback at 1.2V ext .png
    34.3 KB · Views: 1,136
  • QA400 + QA interface  loopback at 7V ext bal div 10.png
    QA400 + QA interface loopback at 7V ext bal div 10.png
    36 KB · Views: 1,083
  • QA400 + QA interface  loopback at 7V ext bal div 10 neg grounded.png
    QA400 + QA interface loopback at 7V ext bal div 10 neg grounded.png
    34.5 KB · Views: 1,060
AX tech editor
Joined 2002
Paid Member
Looks very, very good Demian!
I'm relieved that the 40kHz BW limit isn't - I'd hate to be able to see only harmonics up to a 10kHz fundamental or so.
I guess the droop above 60 or 70kHz could be corrected with a cal file.

I also got from the section on plugins that there's an automatic generation of the measurement data in a text file which is great for publication purposes, though it isn't clear whether that also occurs with 'regular' measurements.

jan
 
Member
Joined 2004
Paid Member
OK, now that it all works and works OK there are some limitations that need to be resolved.

Both the input and the output circuits get distressed when the signal exceeds 8V RMS, even with 17V supplies. Originally I was expecting to get 10V RMS at each point but its not happening. So how to scale?
Option 1 (Current) input divide by 10 or unity, output X16 or X6: problem, both have overload issues withing the operating range of the codec. I could add a resistive divider to correct for the X2 of the output driver and the output won't be clipped when driving a balanced load at 10V but does clip on unbalanced load at 8V.

Option 2, rescale and fix both the output and input to 7V max. Operating range will all be OK but input sensitivity is compromised and the output will need significant attenuation to be linear, however both the DAC and the ADC are better at a reduced level.

Option 3, Rescale output to 7V max, leave fixed. Two settings for the input, unity (1V approx) and -10 dB (3V approx?) or -16 dB (round number near 7 V)?

Making external attenuator cables for input and output is simple and expected to deal with power amplifiers, high gain inputs etc. I want to make it easy to use and hard to get it too wrong with odd gains and different gains that make the on screen arithmetic difficult.

Any suggestions?
 
Member
Joined 2004
Paid Member
its more of an issue of the device under test. If its output is low that you need gain to get it in range of the input of the QA400. if its high (power amp) you need attenuation. Same for the output buffer. The issue is more one of what are you testing? if all I was doing was testing iPods and other mobile stuff (what consumer electronics has become) no gain is needed. To test a traditional preamp or connect to a higher output DAC the output of the DAC is much too high for the QA400.

And then there is the scale factor issue. Its relatively easy to add scaling factors in the software but you need to remember to do it and put the right values in.

Have you gotten your QA400 running?
 
AX tech editor
Joined 2002
Paid Member
Demian how about the age-old 1-2-5 step sequence?
You'll need some more steps but you could go like 50-20-10-5-2-1 V/V

So you would have atten steps 1/50, 1/20, 1/10, then 1/5, 1/2 and 1/1.

For low level sources add gain steps x2 (0.5V/V) and x5 (0.2V/V).

Then you would have an overload capability on each range of some 40% say +3dB.

My unit is still in SF at the USPS sort centre :-(

jan
 
Member
Joined 2004
Paid Member
I was trying to avoid adding the necessary protection stuff and its performance compromises. Not to mention the complexity.

If I fix the output at 7V I get on loopback .0002% THD so the extra gain is not compromising anything. (see below image)

For the input, sticking with the original concept, external probe cables with appropriate connections, a few cables can handle pretty much any input requirements. If I stick with the original X1 input gain and external divide by 10 "probe" would be able to handle pretty much any line level source. If its terminated in RCA or XLR connectors it becomes a complete solution (no more collection of adapters). A separate cable with a divide by 100 ratio and fully balanced could handle almost any power amp. Since the input is fully balanced and the external probes would also be balanced ground loops are not an issue nor are balanced output amps.



If time permits I'll finish a set of cables for this today and try it.

I have also added two pictures showing the "melt noise" feature. It really removes random noise and leaves the deterministic stuff. I suspect the remaining noise is the limits of the 24 bit system. The peak level is about -7.5 dBFS. The signal is scaled, the peak is just below 7V.
 

Attachments

  • QA400 + QA interface  loopback at 7V ext bal div 10.png
    QA400 + QA interface loopback at 7V ext bal div 10.png
    36 KB · Views: 1,020
  • loopback melt noise disabled 6.3V.png
    loopback melt noise disabled 6.3V.png
    33.5 KB · Views: 1,043
  • loopback melt noise 6.3V.png
    loopback melt noise 6.3V.png
    51.7 KB · Views: 374
Disabled Account
Joined 2012
its more of an issue of the device under test. If its output is low that you need gain to get it in range of the input of the QA400. if its high (power amp) you need attenuation. Same for the output buffer. The issue is more one of what are you testing? if all I was doing was testing iPods and other mobile stuff (what consumer electronics has become) no gain is needed. To test a traditional preamp or connect to a higher output DAC the output of the DAC is much too high for the QA400.

And then there is the scale factor issue. Its relatively easy to add scaling factors in the software but you need to remember to do it and put the right values in.

Have you gotten your QA400 running?

My question was one to get you to suggest the answere... which i think you have done.

Most of us would be warry of damaging the front end of the FFt but warnings alone do not always prevent mistakes. I like what Keithley Instruments did on their 823 Nano-Volt amplifier... they put a 'protected' on- off switch on the front panel. can you do same?

Thx-RNMarsh
 
Member
Joined 2004
Paid Member
QA400 measuring Jitter

Here is a jitter measurement using the QA400 Its not scaled the same as the plots on Stereophile, I need to figure that out. The essence is here anyway. The JTest generator is attached. I have not used it in a long time so I don't remember the details.

The jitter is a function of the amplitude of the sidebands of the 11.025 KHz signal. They are very obvious in this plot.
 

Attachments

  • Apogee Mini DAC jitter.png
    Apogee Mini DAC jitter.png
    43.8 KB · Views: 554
  • jitter3.zip
    6.9 KB · Views: 170
Member
Joined 2004
Paid Member
More on Jitter measurement with the QA400

Here is another picture of spectral purity with the QA400. It shows no obvious signs of jitter in its ADC. The Boonton is a crystal stabilized SVO generator that is really good on frequency stability and pretty low noise. The shape of the noise floor probably relates to the disciplining system.

The JTest generator will generate a JTest wave file. I don't have the instructions, but Google could probably find them.

The AP used by Stereophile uses a different sample clock and has a few other differences to the plots will look a little different. However the jitter sidebands relation to the 11.025 KHz reference will be the same and the amplitude of the sidebands translates into Jitter the same way. I have some notes on the amplitude to jitter translation somewhere.
 

Attachments

  • Boonton 48 KHz spectral purity.png
    Boonton 48 KHz spectral purity.png
    31.1 KB · Views: 1,021
I was using Arta with a Behringer U202 external soundcard to try and test a solid state amp I built. Although the noise level was ok, the distortion it had in loopback was about .02%. The good news is that my amp was performing better than the card. The bad news is that I couldn't measure its actual performance. So....I just received my QA400 yesterday. Installed with no issues on an old Dell D600 laptop (1.6 GHz, 2.0 GB ram, OS is XP-SP3). Able to run at 24 bit and 192kHz. Loop back looks good. I hope to do some testing with it today.

The amp I want to test is fairly sensitive, so to drive it I plan to connect it directly to the QA400. However, for measurement, I need to attentuate the amp output by about 20dB and was planning to use a resistive divider to do it. I was planning to use 220 and 2200 ohm resistors. My thinking was that this is large enough so the total load on the amp is small compared to the 8 ohm load it will be running, and the resistance is small enough to keep the noise down and still have a low enough impedance to drive the 100k input impedance. However, in the manual it also says that the AC input impedance is only 10k. So my questions are: Do you think my current path will work, should I select a different resistance for the divider, or should I scrap the idea of a resistive divider and build the interface you have designed?

By the way, the design looks great and thanks for sharing it with us.

Joel
 
Member
Joined 2004
Paid Member
I think 1 Ohm won't be low enough. I don't know the max output voltage of your amp. I suspect its more than 10V. You need to scale as appropriate. I would use a .25 Ohm resistor made from a bunch of paralleled resistors based on what is in the junk box. you may also have a common ground problem between the computer and the amp. That is what stared me on the interface module. if you run the test with the input to the QA400 shorted and tied to the amp ground and still get noise then you have a problem. Running the laptop on battery will help with this. The D600 battery had a pretty short run time as I remember but you should be able to get enough good info.
 
The laptop battery is long gone so I can only run while plugged in. You are right about the ground problem though. I have to have the laptop plugged in with an adaptor to eliminate the ground and allow the QA400 to "float." Found that out with the other sound card.

The amp is a class A solid state, about 10 watts. I designed and built it several years back. It can be driven to about 13 v peaks before it starts to clip. I always thought it sounded great and I am very curious to see how it measures.

I just tried it with a 10 watt .56 ohm power resistor. That scaled the signal nicely. I then tried it with my 220 and 2200 divider. The distortion was quite a bit less with the divider. Now I suspect the .56 ohm resistor was inductive or not "linear" in one other way. I will continue to explore that idea though. I'm also going to make another divider that is more like 13:1.

Joel
 
I connected the two resistors together with a jumper lead, about 14" long. The small resistor is connected to the negative speaker lead, and the large resistor to the positive lead. I measured across the small resistance. Here's a picture of the two components. The larger one specifically says it is non- inductive. I have my doubts about the smaller one.
An externally hosted image should be here but it was not working when we last tested it.

I will try a different component if I can find one.