Trying to build the superbal mixamp

Hi All!

Simulations are great!!, everything works as expected.... in theory... : But when it comes to build it... problems arise...

I have build the D. self, superbal diff input to use it as a mixamp (fixed gain), with four 10k summing resistors, 22k feedback resistors and 2k2 in the servo. a 100R(not shown) and a 47k to gnd in the output and got sound.... sounds good until i realize that the SSM2142 that feeds the balanced summing network, only feeds 2 of the 4 resistors of the summing network. I try with 8 summing resistors, and displaced the wires comming from the outputs of the line driver, to assure i really feed the lower summing resistors. Et voilà! ¡¡silence!!   Something  in the circuit cancels the signal...

Attached are schematics of what i build...
If anyone can spot the mistake/s.....

Thank you for your help!

Jay x

JAY X said:

Hi All!

Simulations are great!!, everything works as expected.... in theory... : But when it comes to build it... problems arise...

I have build the D. self, superbal diff input to use it as a mixamp (fixed gain), with four 10k summing resistors, 22k feedback resistors and 2k2 in the servo. a 100R(not shown) and a 47k to gnd in the output and got sound.... sounds good until i realize that the SSM2142 that feeds the balanced summing network, only feeds 2 of the 4 resistors of the summing network.

Click to expand...

I can't understand a thing from your description.  It seems normal to me that when you connect only one source, it is actually connected to only two resistors!

  I try with 8 summing resistors, and displaced the wires comming from the outputs of the line driver, to assure i really feed the lower summing resistors. Et voilà! ¡¡silence!!   Something  in the circuit cancels the signal... 

Click to expand...

Show us a complete diagram of your circuit and what's connected to it.

JAY X said:

I have build the D. self, superbal diff input ...

Click to expand...

From pg.364 of his book 'small signal audio design', "The capacitor C1 will almost certainly be needed...".

Whatever R31 at the voltage divider might connect to and what the use of R48 might be ....
Probably something at C7 looks missing, whatever this following stage might be ...

Hi!

Attached is a new pdf schematic that show all the stages. : signal-line driver-balanced summing  network-superbal mixamp-unbalanced summing amplifiers- balanced line driver output.

Across the 20k feedback resistor in the superbal diff amp there is a 100pf capacitor.


jay x

Your 'great' simulations look meaningless with shown resistor parts values, as the sim ignores your 10K or 20K labeltexts and calculates with your selected parts values (10R or 20R). These in conjunction with the 100pF caps also result in LPFs set for whishful thinking 80MHz or 160MHz.
From schematic, the DRV134 probably isn't that amused to drive 1.25R (8x10R in parallel) per hi/lo output (not 10Ks and these not ment to be in parallel), and if it could, the summing stage would result in a voltage gain of 8000 or +78dB, not a voltage gain of 8 or +18dB from schematic and your ment to be 10K and 20K parts values or a voltage gain of 1 or +/-0dB with not paralleled 10Ks.
The missing cap should be across R10, not across R9.
Upper part of voltage divider R17/pot probably not 10K.
Leave out R18 (in parallel to the level pot).
Missing resistor between C7 and op-amps inverting input.
Maybe rearange the output buffer for a non-inverting stage in order to have one lesser stage in the audio path and a spare amp for FI a meter buffer or build a servo in order to get rid of C7and C9 ...

Hi!

I really built this circuit, on a strip board, with 10k summing resistors and 22k in the feedback...maybe in the schematic is hard to see...that is why i added text labels...

In any case i will check again after your suggestions.

Jay x.

JAY X said:

Attached is a new pdf schematic that show all the stages.

Click to expand...

You have identical labels (+IN and -IN) on both sides of the 10k resistors - which Spice reads as 10 ohms BTW; Spice considers identical labels are connected together. The output of U6 is thus connected directly to the inputs of U1A. I would think Spice gives funny results.
Also Spice probably reads 2k2 as 2k. 2.2k is correct, 2k2 is wrong.
I'm not sure 100pF is correct; I guess Spice reads that as 100 pico femto Farad. Femto is 10^-15; pico is 10^-12
And again, I don't see the point in connecting the same sourec to all channels simultaneously. I would connect the source to only one input, and leave the others either unconnected (best case in terms of noise) or shorted (worst case).

Hi!

I corrected the schematic as for your suggestions, except the unbalanced summing amplifiers that i must keep as they are.
I built the circuits on a test board, built the superbal with just 2 10k input resistors, instead of all the summing resistor network.

I connected an old Sony discman to a balanced line driver, it output to the superbal balanced input, its output to the unbalanced summing amplifiers, and from them to a balanced line driver to feed my krk v6 monitors...and still can't get a single note out....

jay x

Hi!.

After checking some connections i got sound!!, although very very low. I changed the 20k resistors around the superbal amplifier to 10k. Got more signal but distorted. Then i reduced these resistors to 7k5, and finally to 6k8 with more output level and less distortion. With 6k8 the gain of the superbal stage is 6k8/2*10k = 0.34. -9.37 dbu output level.

This still needs some work...
Jay x

JAY X said:

Hi!

I corrected the schematic as for your suggestions, except the unbalanced summing amplifiers that i must keep as they are.

Click to expand...

On your schemo, teh 2.2k resistor that is supposed to be connected to te output of U1A is not; teh wires cross and are not connected. Same for R14, that goes to R18, but is not connecetd to R16.

U1B and U2B have no power supply connections. Not sure if the simulator you're using knows how a dual op amp works - simply using 'A' and 'B' nomenclature might not be enough to get it to somehow auto-power the second amp of the pair.

Just a guess, since LTspice for example can use a model for a NE5532 but it's really just a single amp. You need to power both explicitly, and powering only section 'A' will not power section 'B'. Not sure which simulator you're using, but it's a 'duh' moment that could save a lot of hassle.

JAY X said:

Hi!.

After checking some connections i got sound!!, although very very low. I changed the 20k resistors around the superbal amplifier to 10k. Got more signal but distorted. Then i reduced these resistors to 7k5, and finally to 6k8 with more output level and less distortion. With 6k8 the gain of the superbal stage is 6k8/2*10k = 0.34. -9.37 dbu output level.

Click to expand...

This is not normal. You must check the DC voltages at all pins of the opamps; they should be about zero, but I suspect there is something that plays havoc with bias, like a resistor not connected to ground or something similar.

JAY X said:

Hi!

I corrected the schematic as for your suggestions, except the unbalanced summing amplifiers that i must keep as they are.
I built the circuits on a test board, built the superbal with just 2 10k input resistors, instead of all the summing resistor network.

I connected an old Sony discman to a balanced line driver, it output to the superbal balanced input, its output to the unbalanced summing amplifiers, and from them to a balanced line driver to feed my krk v6 monitors...and still can't get a single note out....

jay x

Click to expand...

man, the next time you export a bitmap, don't resize it until the component values becomes unreadable.
it's absolutely pointless printing the bitmap as PDF file because you can attach the bitmap file here if you saved it as PNG or JPG.

Hi!

Ok, I will re-send the schematic as picture...