Unexpected gain. help?

some of you might remember my mixer project here: http://www.groupdiy.com/index.php?topic=42359.0

I've built a few units, and I have a wierd problem that I can't explain. some values have been adjusted since then, but the input and output impedances are still the same.
using my NTI MR1 signal generator and ML1 analyzer plugged into the link input and the output passing a -12db signal it comes out at -12db - all good..
BUT if I put two units in series, so MR1 into link1, then out1 into link2 then out2 into ML1 it comes out at -6db  ???

so somehow 0db gain + 0db gain = 6db?

I can't think of any explanation except an impedance mismatch of some sort..
a balancing issue could obviously cause 6db, but it doesn't matter which order I put the two units in, and the mic/aux inputs on the first one in the chain also come out 6db hotter than the second unit

I figure since it's a perfect 6db there has to be something that I'm missing here.

any ideas?

Jason

Looking at what I think may be your schematic or close.

Your link in circuit operates at -6dB, sending to the sum amp and then output stage that makes up for that 6 dB input loss by generating two polarities of the output signal for the + and - outputs delivering + 6db for unity through put.  If one of those two legs is floating and lost between your two units you will see the 6dB loss.

You may also see some weirdness if your test gear is unbalanced, and/or you leave any input or output lines floating or unterminated. I am not familiar with that test gear, so you will need to figure out exactly where your test signal is being lost. On the input side coming into the + or - input one legged (and floating the other) will give different results (besides the polarity flip).  One the output side you need to capture both legs to get full output level.  While it seems this would show up in testing a single unit.

I vote for an open leg in the link cable between the two units.

  JR

JohnRoberts said:

I vote for an open leg in the link cable between the two units.

  JR

Click to expand...

I would have suspected that if I lost 6db along the way, but an accidental unbalancing shouldn't cause an extra 6db when I add an extra unit and a link cable.
I've even tried the units in the other order, and the exact same results.

I can also rule out my signal generator because if I use the aux inputs on each unit the gain is different to the final output.
and the ML1 tells me that the output is evenly balanced. (plus just listening to it through a preamp I hear the same thing that I'm measuring..

I'll try another link cable just for fun but I don't expect any changes.
ps: I also triple checked the connector pinouts and everything looks good.

Jason

ok, answering my own question here, after a bunch of frantic testing..

turns out the cable from my signal generator to the first unit had a bad leg, which I never caught because the DUT was balancing the signal and it came out the other end just fine..
problem is, I added gain to the circuit to make the total gain 0 (without double checking what the gain SHOULD be with those values) and inadvertently made the gain of each unit 6db (to make my meter read 0)
and unfortunately the MR1 will say on the screen that it's putting out a -12db signal even though the unbalanced signal at the other end of the cable is actually -18
now that I've fixed the cable I get a gain of 12 through the two units, and 6 through each of them independently.. so all is solved. (just have to change some values now)

thanks!

As I mentioned if you come into the link input single legged, you will get different results depending on which input you use (if the other input is not terminated with a low impedance). Coming into the opamp - input (indicated + input on schematic) you get the expected -6dB gain. OTOH, coming into the opamp + (indicated -), with the other input floated will give roughly -10dB gain... or 4 dB less than expected. If that other input is grounded or connected to a low source impedance the gain comes up to the expected -6 dB again.

Simple dual driven opposite polarity outputs will not deliver + 6dB if both are not connected.

It's always something...   BTW the schematic looks like it should deliver unity gain from link inputs to output with those values.

JR

right. and I built my prototype with that unity gain setup and when I measured it came up 6db short (because of my bad wire), so instead of asking why, I added gain.
and I didn't realize the problem until I made a run of them and didn't get what I expected.

live and learn.

Jason

You must learn to trust the force (math) Luke, and trust your test equipment a little less when it doesn't agree with the force...

I guess you could always make up a bunch of bad cables....  :

JR

since you solved your problem, i can fire my question  ;D

a while ago, i was very attempted with those test equipment you have,
are you happy with them, i am very keen on getting mr2 and ml1 too...
is ml1's display clear/large enough to read some graph ?
any comments

thanks

kambo said:

since you solved your problem, i can fire my question  ;D

a while ago, i was very attempted with those test equipment you have,
are you happy with them, i am very keen on getting mr2 and ml1 too...
is ml1's display clear/large enough to read some graph ?
any comments

thanks

Click to expand...

the MR1 is a great tool, and many people seem to have one in their toolbox. although as I've come to find out, the display only tells you the output level of its balanced signal, if you choose to (or if you accidentally) unbalance the signal the actual level won't be what the display tells you. not a big issue, just something to be aware of.
I think for a generator that does tones, pink/white noise, polarity test (with ML1) and more, it's a good value.

the ML1 is very handy in the field. using the pair it is very quick to diagnose common problems. run pink through the system and then plug the ML1 into every intermediary point until you find the signal change..
as a bench tool though, I've found a USB audio interface and software like SMAART and similar to give a more useful readout. the screen on the ML1 isn't exactly HD, and although with the USB upgrade (or with an AL1) you can connect it to your computer, the display on the computer is an exact replica of the ML1 (but with the benefit of data logging)

but for instance, for this mixer I'm building the ML1 can tell me what level the noise floor is at, but I need a good FFT analyzer like SMAART to tell me if the dominant noise is a hum or a hiss (both of which have different solutions)
so I use the NTI combo to quickly tell me what my gain structure is like, but not to take qualitative measurements.
however, if I was headed to a service call for an existing (previously working) system, the NTI kit has all that I need.

Jason

many thanks for the detailed info 
kambo

I designed the old Loftech TS-1 so I have always wondered what it would be like to revisit that with modern technology.

JLavoie said:

the MR1 is a great tool, and many people seem to have one in their toolbox. although as I've come to find out, the display only tells you the output level of its balanced signal, if you choose to (or if you accidentally) unbalance the signal the actual level won't be what the display tells you. not a big issue, just something to be aware of.
I think for a generator that does tones, pink/white noise, polarity test (with ML1) and more, it's a good value.

Click to expand...

A simple test would be an optional operator selection to single end or differential the test equipment output... Perhaps with gain correction (usable at modest output levels).

If the return level changes when single legged, you know you aren't correctly reading both legs, either going or coming back.

I suspect the differential output is used mainly to get decent drive level from relatively low voltage battery power.

the ML1 is very handy in the field. using the pair it is very quick to diagnose common problems. run pink through the system and then plug the ML1 into every intermediary point until you find the signal change..
as a bench tool though, I've found a USB audio interface and software like SMAART and similar to give a more useful readout. the screen on the ML1 isn't exactly HD, and although with the USB upgrade (or with an AL1) you can connect it to your computer, the display on the computer is an exact replica of the ML1 (but with the benefit of data logging)

Click to expand...

With modern chips you can probably do simple FFT for only a few dollars... Difficulty is making the output meaningful for different measurements may be the hard part. 

but for instance, for this mixer I'm building the ML1 can tell me what level the noise floor is at, but I need a good FFT analyzer like SMAART to tell me if the dominant noise is a hum or a hiss (both of which have different solutions)
so I use the NTI combo to quickly tell me what my gain structure is like, but not to take qualitative measurements.
however, if I was headed to a service call for an existing (previously working) system, the NTI kit has all that I need.

Jason

Click to expand...

Yes, you can do most troubleshooting with far lower resolution than bench grade test gear.

JR