[julienvoirin]

dear boxers
i've got a hard problem
I've made the Aout-ng module of smash tv, in a midibox cv.

I use a bipolar PSU -12V + 12v to supply aout ng on J3 (+12, gnd , -12v). i supply the core (completly stuffed) at J1 with gnd and +12V pin of the bipolar PSU.

When in unipolar mode, i have the expected range (from zero to 10,67V). it's how i calibrated the gain before the opamp, using P1.
Then i switched to bipolar mode, and set P9 (on channel CV1) to adjust the "middle" (in mb cv menu). I well have zero volt.

BUT, when i set "min" and "max" i only have -2,76V and +2,76V. I should have -5V, +5V.
I have this problem on all the channels.

If i remove the gnd from Aoutng J3, and supply the core with a wallmart (9V DC) , i got a bias of +10V on my min and max voltages in bipolar ( min = 7V, middle= 10V, max = 12,56V ) !
The ground of Core is connected to the ground of AoutNG.

So what's the problem ???

[seppoman]

julienvoirin, on 30 July 2010 - 01:44, said:

If i remove the gnd from Aoutng J3, and supply the core with a wallmart (9V DC) , i got a bias of +10V on my min and max voltages in bipolar ( min = 7V, middle= 10V, max = 12,56V ) !
The ground of Core is connected to the ground of AoutNG.

removing the GND doesn't work because in this setup, the +/- 12 V have no defined relation to the GND of the other PSU. If you want to use two separate PSUs, you always need to make a (single) connection between both GNDs or strange things will happen (as you've observed).

S

[julienvoirin]

thanks for your explanations seppo. it is what i thought

so i use one single PSU (-11,96V, Gnd, +11,8V) to supply Core on J1 (with +11,8V and Gnd) and AoutNG on J3.

Here are the voltages measured on free leg of the resistor Ri (black probe connected to Gnd of J4 on Aout), jumpers set on channel1 as bipolar (others channels not set) :

Min (V/oct) :
CV1 out = -2,78V
R41 : -2,77V
R33 : -1,43V

R25 : -1,05V
R17 : -1,05V

Middle (V/oct):
CV1 out = 0,00V
R41 : -2,69
R33 : -1,34

R25 : 0,00
R17 : 0,00

Max (V/Oct) :
CV1 out = +2,78V
R41 : -2,60
R33 : -1,26

R25 : +1,05
R17 : +1,05


In unipolar i well have the range (0V ; +10,67V ) as expected

I only connected together the PSU, the Core, a small control surface (just 74hc165 + 1 encoder + 2 buttons + C 100n) and the Aout NG modules.
Gnd of core, psu and aout are connected together (tested with a multimeter)

This post has been edited by julienvoirin: 30 July 2010 - 15:24

[seppoman]

Hmm..

well the R33/R41 measurements don't mean much as you only can measure the voltages that are the "rails" of the trimpot. I suppose if you measure these values, you mean you are probing the open resistor pins on the top? R17/R25 having the same values also isn't a surprise as these legs go to the same PCB trace

But -1..+1V are a bit strange cause that looks like the DAC does only half the usual swing.

Can you please remove the jumpers altogether and then measure the very first bottom one of the pins (or "R17/25", that's where they go)? This should give you about -2V and should change if you adjust P9. After that, please measure the upper right pin of this jumper pair, that's the one where the trace coming from the DAC is going to, and again set the box to output min and max value. This should be 0...4V.

And you can check if R17 and R25 are both 10k ones (brown black orange) and there's no short or bad soldering around these or the trimpot.

S

[julienvoirin]

Quote

measure the very first bottom one of the pins (or "R17/25", that's where they go)? This should give you about -2V and should change if you adjust P9

it does ; from -2,75 to -1,51 while adjusting P9

Quote

DAC is going to, and again set the box to output min and max value. This should be 0...4V.

min =0v
middle = 2,03v
max = 4,06v
measured directly on the DAC and JP1

Quote

R17 and R25 are both 10k ones (brown black orange)

it is, but brown black black red. measured value = 9,96k
really lost

i am testing shorts now

[julienvoirin]

seems there is no shorts
i got this prblem on 2 aout ng pcb from smah tv. maybe there is defects into the pcb (!) as it could have been a bad batch
i start to be disappointed as i wanted to use he ssm filters

could it be software related ? i recompiled it

This post has been edited by julienvoirin: 30 July 2010 - 16:51

[seppoman]

hmmm. I don't really have an idea what's wrong here. That the unipolar mode works fine shows that the amplifying stage and DAC work fine. The last results show that the offset generation is ok. DAC alone works... The problem can't be on the software side if uni mode is fine, then you change jumpers (and nothing else, i.e. also no changes to software settings etc), and then it doesn't work... I can't imagine what kind of PCB error would lead to this behaviour, and the chances for one person to get more than one PCB with the same kind of error on all 8 outputs is virtually nil.

One other measurement you could do is to place only the right jumper (the one where the DAC trace goes to) and leave the left one empty and measure min/max on the left bottom pin (or R17/25). This should give you roughly -2..2V depending on the pot setting.

S

[julienvoirin]

Quote

This should give you roughly -2..2V depending on the pot setting

-0,72 to -1,36V
a track to follow ?

[seppoman]

julienvoirin, on 31 July 2010 - 08:49, said:

-0,72 to -1,36V
a track to follow ?

I meant to measure there with min and max setting in software, this sounds like you had it on min and measured the extreme settings of the pot? If that's the case, this would be again the -1 V for min, if not, I can't make much sense of these readings yet.

S

[julienvoirin]

-1,06V to +1,06V
As the Gain of TL074 is something like 2,5 (4,06V x G = 10,67) it is logical i got -2,76 ... +2,76V
So now my question : how can I get a -2 +2 voltage on the P9 circuit ?? My supply is +/-12V
Isn't there a problem of gnd and relative ground IYO (for the core) ?

[seppoman]

Ok, so you can finally check that there's no connection between the left two pins while the jumper is away, then we can be sure that we can leave the gain/output stage out of the further troubleshooting.

As you've tested that the two GNDs of both PCBs are connected, it can't be a GND issue. And even a lower "+/-12V" supply could only cause a wrong offset, not a wrong voltage swing at this point.

At the moment I'm out of new ideas - one thing you could do is take a few decent high resolution pictures of both sides of your AOUT_NG and of your whole setup, maybe I/we could see something suspicious.

S

[julienvoirin]

Quote

there's no connection between the left two pins while the jumper is away

no connect

Quote

it can't be a GND issue

we both agree

why measured values on AOUT PCB of R41 is 0,99k ? R33 1,1k ?
I measured the same resistors from Reichelt bag (same colour code) and there are the expected values


btw, thanks a lot for helping me to debug, really nice

i tested various things : changing GND reference (from the PSU, from the core), removing power supply leds (supply proof to check there is well -/+12v), testing without core, etc ... nothing change, except a bias as you mentioned earlier (and it is logical)
without core i well measure -2V on R17/25 (DAC 0V, Min), but i can't know if there is +2V at max as the core is deconnected (Max value, DAC @ 4V)

[Shuriken]

julienvoirin, on 31 July 2010 - 12:22, said:

no connect
we both agree

why measured values on AOUT PCB of R41 is 0,99k ? R33 1,1k ?
I measured the same resistors from Reichelt bag (same colour code) and there are the expected values

As i have two untested AOUT_NG laying here, i am following this with great interest. R41 should be 1.5k and R33 should be 2.2k. So i assume you measured those soldered on the board?


I measured R33 en R41 on one of the boards and got the same values.

This post has been edited by Shuriken: 01 August 2010 - 08:33

[lylehaze]

Quote

i supply the core (completly stuffed) at J1 with gnd and +12V pin of the bipolar PSU.

OK, so I really shouldn't jump in so late in a conversation.. but I have a single observation.
Based on the above quote from the first post, there may be a grounding problem after all.

J1 of the (pic) core does NOT have a ground connection. It only feeds a bridge rectifier.
If you are feeding DC to power the core, and you need both boards to have the same ground,
it seems a better idea to connect the incoming DC to the + and - corners of the bridge, instead of
using J1.

If I'm wrong, please correct me.. but if I understand that quote correctly, there will always
be a 1 diode drop between the two grounds, which may cause funny problems.

Good Luck,
LyleHaze

[julienvoirin]

Dear boxers
I am very happy to get different opinions on my problem

So, i have 2 midiboxes with each an Aout NG : Seq V4 and MB CV

The seq V4 has a power supply like this : -12V, Gnd, +12V, +5V. So the same GND.
It is a commercial ELKA PSU for computers, but it isn't a switched PSU. 78xx share the same cooler, 7912 is on another.
+/-12V supply J3:Aout and the GND is brought by the J19:core32 to J1:Aout NG
Gnd and +5V supply J2:core32 (so directly without diodes bridge). But the bridge is soldered on the Core32 .. USB power option is not enabled.

The MBCV has only one bipolar PSU : -12V, Gnd, +12V directly connected to the Aout NG J3
To supply the core, i connected wires between Gnd and +12V of the PSU to J1:Core
Core and Aout are connected together by a ribbon cable between J7:Core and J1:AOUT. Normally, i've set the Gnd wire ; i will check tomorrow.

in both case i don't have the full range in Aout bipolar mode : (-2.76 ; +2.76)

When I measure the voltages between the PSU Gnd and all the different points were Gnd is supposed to be, i've always got 0.00V, on both midiboxes, core or Aout modules. +5V exists were expected, +/-12V too.
I will put pictures tomorrow online (tonite i am at work )

This post has been edited by julienvoirin: 01 August 2010 - 22:10

[vcfool]

Well, so I have the same situation here.

I have an AOUT_NG inside a MB-SEQ v4. I have been using Unipolar mode without any problems since I finished it some months ago. I have been following this topic and just now I tried the Bipolar mode.

So far I have only tried CV1: On the Mixer Map I assigned a CC to the AOUT CV1 output and calibrated P9 to have 0V at a CC value of 064. And I have the same problem: the output range is only -2,7V to +2,6V.

[seppoman]

Oh. Uhm. Ah.

F.u.c.k.

Mine also gives -2.7...+2.7V - Actually it seems I was simply assuming the -5..+5V thing and never really measured it. I was satisfied that the needs of the SSM2044 are fulfilled. Actually, as the NG is out already more than two years, it seems nobody else ever stumbled upen this...

I need to think about this for a minute... Changing the gain for the bipolar channels isn't much of a problem, still I'd like to analyze why this is so before I'll give an advice.

Sorry for the inconvenience, this is really embarrassing

S

[nILS]

[julienvoirin]

me -> <- seppo
at least I know i worked well

the best is certainly to only change R33/41

This post has been edited by julienvoirin: 02 August 2010 - 22:19

[seppoman]

Ok solved... if you sum two voltages through two resistors of same value, you'll get (V1+V2) / 2 as a result.... stupid me...

There are two solutions to this problem:

The first one (that I'd prefer) is to change the value of R9...R16 to 5.6k, only on the channels that are setup for bipolar mode. This will give you a calibration range of about (+/-) 3.5V to 5.8V. That should be sufficient as a hot fix, isn't too much effort and you'll still have a decent precision when calibrating.

The second one: you could exchange R1...R8 for 10k and R9...R16 for 2.2k (on all channels). That would give you a calibration range in unipolar mode from 9.5V to (22V) the maximum the TL074 can give (a bit more than 11V). And the range for bipolar mode would be about (+/-) 2.4V to 5.5V. This solution would preserve the "just change the jumpers if you need some channel in bipolar mode" thought behind this option. On the other hand, it would dramatically increase the calibration range, i.e. exactly calibrating the outputs will get harder because the same angle of turning the pot will have much more impact. So although this will lead to more documentation effort and questions, I'd still prefer to generally suggest the first solution instead of a) having Smash change the contents of his kits and b) making the module harder to handle for the 95% of people that won't ever need the bipolar mode

I can only apologize for the confusion and this stupid design mistake... Wiki will be updated in the next days.

BTW to all SSM2044 builders: The module is designed with this (wrong) bipolar range, so you don't need to change anything on your AOUT_NGs. I guess for proper +/- 5V operation I'll have to try out a few other resistor values on the SSM module, but this will only apply if you if you're using a different CV source than an AOUT_NG.

And BTW2: I've just had a short look at the old/original AOUT, and the bipolar mod is the same concept like what I'm doing on the NG (guess that's where I got it from ;)), so this module will probably show the same behaviour.

S