Test & Measurement interface for Soundcard

Thank you for the explanation grufti. So the blue line plot is the SCI+soundcard and red line plot is the soundcard alone? If that's true, I would be worried because the blue plot seems to have more artifacts than the red plot, at 60Hz, 2kHz, and 3kHz. Also, have you in any way shifted the red plot lower or the blue plot higher, or this is how they came out of the analyzer? If you did not shift them, it would seem the soundcard alone has significantly lower noise floor. Did I get this wrong?

What soundcard is it, by the way?


Indeed, red line soundcard, blue line sound card plus sound card interface at the settings and levels I listed in the previous message.

The soundcard is not really a soundcard. It is an external Lynx Aurora 16 ADC/DAC in combination with a Lynx AES16 interface.

I would have been extremely surprised, if my build of Pete's interface matched the Lynx combination. I'm extremely pleased that the interface behaves as well as it does.

You can ignore the absolute level of the red soundcard line and with it the level shift between the two results. All I wanted to do with this test was to get near optimal performance out of Pete's sound card interface.
 
Last edited:
I've always wanted an AP, and have used the 2722 in ATE qualification work (ac front end for an instrument in a semiconductor ATE test system). I would settle for a System One single or dual domain, but even these discontinued models sell for a hefty chunk of change on eBay and elsewhere. So this answers my needs too..

Keep your eyes out for a Boonton 1120 -- the generator is close in performance to the earlier AP's, the a.c. meter is very, very good. The other one to look for is the Tektronix SG5010/AA5001. Agilent and Stanford Research have also introduced new models which compete well.
 
Administrator
Joined 2004
Paid Member
Keep your eyes out for a Boonton 1120 -- the generator is close in performance to the earlier AP's, the a.c. meter is very, very good. The other one to look for is the Tektronix SG5010/AA5001. Agilent and Stanford Research have also introduced new models which compete well.

Hi Jack,
Sounds like something to look for at spring hamfests, thanks for the tips. :D
 
Hi,

My DIY hobby has mostly concentrated to speaker building, but now I've become very interested of amplifier building too. I just ordered three books about amplifiers and electronics to study and now I need to start to make somekind of measuring equipment. I've actually built the Millett Hybrid headphone amplifier few years ago, but that's pretty much all I've done with electronics.
I've got a M-Audio FW410 soundcard for measurements which is pretty good and I've also bought SoundEasy software which seems to have oscilloscope/spectrum analyzer functions. I've also thinking of buying dummy loads from parts express so I can make measurements to different loads and try amplifier limits.

This Millett measurement interface could just fit my needs perfectly, but as I really am a total beginner with these equipment, I need to ask some dumb questions about the very basic connections.

-I don't really understand the "gen output" connection, If I want to measure an amplifier in stereo load, can I just put cables straight from my soundcard to the amplifier inputs?

-How should I wire the bnc input to the amplifier outputs? Do I just need a single wire from the amplifier output (driving a dummyload) to the bnc connector pin? I'm always confused about the grounding connections.

-I've only got a cheap digital multimeter, so could someone clear for me how the 50hz signal calibration is actually done for this device?

I would also like to ask you about different solders. What I've used is 60/40 0.8mm solderwire, but I read from tangentsoft website that the 63/37 might just be better to use. I noticed that there are quite a lot of different solders available:

http://fi.mouser.com/catalog/640/2148.pdf

What would you recommend, and what is the easiest to use? The "245" seems best to me, but I don't know anything about the flux percentages.

If I have to remove an earlier soldered wire or component, the solder doesn't seem to melt that easily (or at all), I believe this is because the flux has burnt away? What sould I do in this kind of situation? Do I need a somekind of flux to make the solder melt again?

I also sometimes solder quite big connectors and it's always hard to get big amounts of solder to melt nicely, what type of solder is best for big connectors?

-Ari S
 
Administrator
Joined 2004
Paid Member
Tonight I found out it is a really good idea to ground the shaft/body of the rotary switch in order to eliminate pick up of 60Hz and its harmonics. I noticed moving my hand closer to the metal knob on this switch resulted in a >20dB increase in 60Hz spectra and the addition of somewhat lower level harmonics at 120Hz, 180Hz, and 240Hz.. I grounded the switch with a piece of 28ga wire between the lock washer and the body of the switch and grounded it to the board ground. This totally eliminated any pick up with no noticeable line frequency or related spectra. (My case is totally shielded on the top and bottom, but not sides, front or rear - and does not appear necessary to do so.)

The 24192 has significant 2nd and 3rd harmonic which is definitely generated in the sound card - a lot of it actually seems to come from the dac, but some is also generated by the adc. Overall 2nd harmonic is -105dBr and the 3rd about -110dBr relative to the fundamental (1kHz). The op-amps and the INA/DRV amps probably are not significantly better according to their nominal specs.

The other interesting thing I have noticed in my unit is that 2nd harmonic increases by nearly 20dB when I set the generator to balanced output mode as compared to unbalanced mode - wondering what is up with that. I will investigate over time.

I have been talking to Ulrich and have found some pretty consistent ways to crash audiotester so I will be sending him some screen captures of the error messages - hopefully this will help to resolve a few issues.
 
Last edited:
Administrator
Joined 2004
Paid Member
Quoted from Twisted85

-I don't really understand the "gen output" connection, If I want to measure an amplifier in stereo load, can I just put cables straight from my soundcard to the amplifier inputs?

The "gen out" connection provides a buffered low impedance (high current) balanced or unbalanced output. The sound outputs of your sound card have limited drive capability and can be degraded by certain types of loads connected to them. In the event of a malfunction in the device you are testing your sound card is much less likely to be damaged.

-How should I wire the bnc input to the amplifier outputs? Do I just need a single wire from the amplifier output (driving a dummyload) to the bnc connector pin? I'm always confused about the grounding connections.

Both the pin and the shell must be connected, the input can be floating or ground referenced as needed, but the input must be referenced to the amplifier output common or negative output (gnd mode or floating) or in the case of a bridge output type amp you would connect the negative output to the shell and use floating mode.

-I've only got a cheap digital multimeter, so could someone clear for me how the 50hz signal calibration is actually done for this device?

The calibration is handled primarily by the software, and verification with a meter with good AC accuracy is recommended even with 50Hz or 60Hz tones. Specific instructions will be found in the documentation for your software under the calibration heading.


I don't use RoHS compliant solder as we are not yet required to do so for personal use, and I have quite a lot of lead bearing solder left. (I hate soldering with lead free solder - even so it really helps to use an external liquid flux and not rely on the flux in the solder..)

I generally use 60/40 Kester or Wonder Solder (specific formulation is a matter of some conjecture.. )

WRT to your comments about reheating joints to remove unwanted components it sounds to me like your iron is not hot enough, Pb free solders require somewhat higher heat.
 
Last edited:
Tonight I found out it is a really good idea to ground the shaft/body of the rotary switch in order to eliminate pick up of 60Hz and its harmonics. I noticed moving my hand closer to the metal knob on this switch resulted in a >20dB increase in 60Hz spectra and the addition of somewhat lower level harmonics at 120Hz, 180Hz, and 240Hz.. I grounded the switch with a piece of 28ga wire between the lock washer and the body of the switch and grounded it to the board ground. This totally eliminated any pick up with no noticeable line frequency or related spectra. (My case is totally shielded on the top and bottom, but not sides, front or rear - and does not appear necessary to do so.)

<SNIP>

The other interesting thing I have noticed in my unit is that 2nd harmonic increases by nearly 20dB when I set the generator to balanced output mode as compared to unbalanced mode - wondering what is up with that. I will investigate over time.

I did mention the "steel" cookie tin, did I not?

For grounding the chassis, manufacturers like Agilent, Tek, Boonton use what appears to be an Emerson Electric uninsulated binding post with knurled thumb -- DK has it as part number J587-ND. You could just as easily use a Radio Shack banana receptacle. You could use the green #12 wire available in the electrical aisle at Home Depot or Lowes -- but it's better to just gut one of the orange extension cords as the green is much more flexible -- and it's cheaper as well.

Cadmium in solder is not necessarily a "bad thing" -- less stress on components from lower melting point, higher strength, greater wear resistance and an established recycling mechanism. Cadmium poisoning generally comes from inexpensive costume jewelry. Mercury in light bulbs is a "bad thing" as folks will just chuck them into the garbage.
 
Administrator
Joined 2004
Paid Member
I did mention the "steel" cookie tin, did I not?
<snip>

Indeed you did, and I fully understand the ramifications of that comment. Like many here I chose the plastic case - so I thought I would share my experiences with shielding and noise pick up issues in the hope of saving someone else the aggravation. (Particularly those less experienced)

I knew in advance that I would shield the case, but did not stop to think about the fact that the switch shaft actually wasn't grounded. The result of that omission was predictable to you or I, but perhaps not to everyone participating in this thread.

The ground jack suggestion is an excellent one, and I think I will eventually add it. I'm also planning on installing an LED to indicate when the thing is on since I did not build it with the meter.

I've had a lot of fun putting this box together, and now it is time to start using it.. :D
 
Hello All,
I agree with all of it. This Millett Interface DIY build is educational including the shielding, voltage dividers, insights and feelings of the fellow builders. Thank you.
My thoughts. I teach soldering, brazing and welding at my night job. Many of the metals we DIYers contact daily may be toxic. Even the screws and nuts (do not put them in your mouth) may be cadmium plated. Yes the solder contains lead. Even the copper pipe we drink out of is suspect. Blah Blah Blah.
Check out the links below to info about Heavy Metal Toxicity and a solder Material Safety Data Sheet.

Heavy Metal Toxicity: Online Reference for Health Concerns
http://www.bemidjistate.edu/offices/environmental_health_safety/msds/sldr_kstr.pdf
DT
All just for fun! Don’t pick your nose!
 
thosuk

I see you are also in the UK, have you sourced your components locally to build the interface, I had a quick look at Farnell and could find a good few parts though some had minimum order quantities, if I order in quantity I may have some surplus, if you are interested let me know

Stuart
 
Member
Joined 2005
Paid Member
405 Man
Thanks very much for your kind offer.
I already took the easy way and just used the DigiKey and Mouser route.
There were a few min order issues but only for low cost components.
Probably similar components to your likely surplus :)
Assuming they actually deliver, I too will be happy to help out anyone building this interface.
Tom