Author |
Message |
Jaba
Joined: Feb 27, 2009 Posts: 48 Location: Genova, Italy
|
Posted: Mon Jun 13, 2011 8:24 am Post subject:
Balanced Modulator troubles |
|
|
Hi,
I just found out that my Yusynth Balanced Modulator does not behave like it should:
when I use the AC coupled inputs it is ok, the outputs are a bit low but they behave like I expected,
but with DC inputs it does something strange.
After some tests it seems like it has fixed offset voltages on all inputs.
I tried connecting a scope to the DC inputs, without any signal, and I saw several volts DC (about 3.6V and 7.3V); I think this is not normal, is it ?
I thought of some wrong wiring, wrong resistor values, ..., but I found none yet.
any help ?
thank you
Paolo(Jaba) |
|
Back to top
|
|
|
yusynth
Joined: Nov 24, 2005 Posts: 1314 Location: France
|
Posted: Tue Jun 14, 2011 4:48 am Post subject:
|
|
|
I'll check that. By the way if you short-cut the DC inputs to ground do you still have offsets ? _________________ Yves |
|
Back to top
|
|
|
Jaba
Joined: Feb 27, 2009 Posts: 48 Location: Genova, Italy
|
Posted: Wed Jun 15, 2011 2:15 am Post subject:
|
|
|
Hi Yves,
I am not sure that I understand what you wrote
If I ground an input, then of course it is at 0V, i.e. no offset
If I leave a DC input open and I send a signal to the other AC input, the output is 0V; this is actually the same configuration used for calibration, so it is obvious, i.e. the trimmers were actually set to obtain this zero output
But then if I send the signal to the AC input and ground the other DC input, then I get a strong output; this should not be, if I understand correctly
I tried the above with all different combinations (AC A with DC B, AC B with DC A, AC C with DC D, AC D with DC C), they all gave the same result, with different amplitudes
Using both AC inputs all seems ok, the output is as expected i.e. the multiplication of the AC components of the two input signals
I still don't understand, maybe in the end I should look for mistakes once again
Jaba(Paolo) |
|
Back to top
|
|
|
yusynth
Joined: Nov 24, 2005 Posts: 1314 Location: France
|
Posted: Wed Jun 15, 2011 2:51 am Post subject:
|
|
|
I mean if you short-cut to ground both DC inputs (A and B), do you still have an offset at the DC output (AxB)?
Same for (C and D) DC inputs and (CxD) DC output
By the way did you follow the trimming instructions on my page ? _________________ Yves |
|
Back to top
|
|
|
Sebo
Joined: Apr 27, 2007 Posts: 564 Location: Argentina
|
Posted: Wed Jun 15, 2011 3:24 pm Post subject:
|
|
|
Hi Yves:
I did measure the same thing in mine, I followed the calibration procedure, but seeing the schematic I think that some of the voltage set by the trimmers to compensate the offset will get back to the DC inputs. Am I right? _________________ Sebo
---------------------------------------
My Music:
https://www.facebook.com/cosaquitos/ |
|
Back to top
|
|
|
Jaba
Joined: Feb 27, 2009 Posts: 48 Location: Genova, Italy
|
Posted: Thu Jun 16, 2011 3:32 am Post subject:
|
|
|
Hi Yves and Sebo,
- yes, if I ground both DC inputs I get about -7V on the DC output, tried on both modulators
- yes, of course I followed your trimming instructions, that is what I referred to as "calibration" in my previous message
This Balanced Modulator is almost the same circuit as the Formant ring modulator, but the Formant version has no DC inputs; maybe there is a reason for this ... hmmm
Yves, do you measure similar voltages on your unit too ?
Jaba (Paolo) |
|
Back to top
|
|
|
yusynth
Joined: Nov 24, 2005 Posts: 1314 Location: France
|
Posted: Thu Jun 16, 2011 3:51 am Post subject:
|
|
|
Jaba wrote: | Hi Yves and Sebo,
- yes, if I ground both DC inputs I get about -7V on the DC output, tried on both modulators
- yes, of course I followed your trimming instructions, that is what I referred to as "calibration" in my previous message
This Balanced Modulator is almost the same circuit as the Formant ring modulator, but the Formant version has no DC inputs; maybe there is a reason for this ... hmmm
Yves, do you measure similar voltages on your unit too ?
Jaba (Paolo) |
Ok I'll do check this tonight. If there is such a constant bias at the DC output this can be corrected using a resistor connected to the +15V rail on one side and connected to the non-inverting input of the OPamp.
Did you use 1% resistors for R11,R19,R20,R21,R33,R41,R42,R43 and R17,R18,R39,R40 ? _________________ Yves |
|
Back to top
|
|
|
Jaba
Joined: Feb 27, 2009 Posts: 48 Location: Genova, Italy
|
Posted: Thu Jun 16, 2011 5:40 am Post subject:
|
|
|
Yes, I used all 1% metal film resistors |
|
Back to top
|
|
|
yusynth
Joined: Nov 24, 2005 Posts: 1314 Location: France
|
Posted: Thu Jun 16, 2011 8:40 am Post subject:
|
|
|
OK I made the measurments with all the inputs unconnected (left alone so to speak) and I obtained on the DC output for the two balanced modulator :
178mV and - 80mV respectively.
Therefore there is a slight offset (the MC1496 is a cheap IC compared to an AD633 which would provide a smaller offset) but this is far from the value you obtain.
Can you post a photo of your PCB, component side and track side ? _________________ Yves |
|
Back to top
|
|
|
yusynth
Joined: Nov 24, 2005 Posts: 1314 Location: France
|
Posted: Thu Jun 16, 2011 8:49 am Post subject:
|
|
|
Sebo wrote: | Hi Yves:
I did measure the same thing in mine, I followed the calibration procedure, but seeing the schematic I think that some of the voltage set by the trimmers to compensate the offset will get back to the DC inputs. Am I right? |
Yes and compensate for it. Normally all is cancelled, there should be no offset, the differential output is meant to eliminate the offset. However the MC1496 is a simple balanced modulator designed for RF not DC applications and it cannot be as performant as a AD633 which cost about 60 times the price of a MC1496. In order to compensate for the residual offset one can add a trimmer (connected between PSU rail) and feeding a resistor connected to one of the input of the first TL072 _________________ Yves |
|
Back to top
|
|
|
Jaba
Joined: Feb 27, 2009 Posts: 48 Location: Genova, Italy
|
Posted: Thu Jun 16, 2011 8:54 am Post subject:
|
|
|
ok, maybe I was not clear enough:
- if all inputs are left unconnected or if I only use the AC inputs, then all is working ok, no ugly offsets, both modulators are quite good and useable
- but if I connect the DC inputs, then the trouble starts, with offsets
- if I short to ground one or both the DC inputs, it behaves as if a DC offset was instead of the 0V at the input
what do you measure at DC output when you ground both DC inputs ? |
|
Back to top
|
|
|
yusynth
Joined: Nov 24, 2005 Posts: 1314 Location: France
|
Posted: Thu Jun 16, 2011 9:13 am Post subject:
|
|
|
When I ground the DC inputs (connected to zero) I also obtain -7V at the DC output. That's weird indeed. I'll check the datasheet to understand... _________________ Yves |
|
Back to top
|
|
|
Sebo
Joined: Apr 27, 2007 Posts: 564 Location: Argentina
|
Posted: Thu Jun 16, 2011 2:26 pm Post subject:
|
|
|
What I mean is, if I didn't connect anything to the module I measure -4V in one DC input and +7V in the other one (both modulators measures the same).
So I think this voltages are the compensation for the offsets, when you ground a DC input the offset is gone, and that's reflected at the output. _________________ Sebo
---------------------------------------
My Music:
https://www.facebook.com/cosaquitos/ |
|
Back to top
|
|
|
Jaba
Joined: Feb 27, 2009 Posts: 48 Location: Genova, Italy
|
Posted: Tue Jun 21, 2011 1:42 am Post subject:
|
|
|
Hi,
after some experiments, I have found a working solution without any apparent drawback, so I think it would be nice to share it.
Here it goes:
I added a voltage follower before each input (that is one for A input, another for B input, ... a total of 4, each one serving both AC and DC coupled inputs) and I added four 180k resistors: from pin 8 of each LM1496 to GND and from pin 1 of each LM1496 to GND.
some more details:
the 1microFarad input decoupling capacitors were mounted on the input jacks; I left them there, added a 100k resistor from each DC input to ground and then disconnected from the PCB the wires going from the jacks to the PCB; now those wires go to the voltage followers (on a small perfboard) and the outputs of the followers then go to the original PCB inputs.
For the voltage followers I just used common opamps, with their inverting inputs directly wired to their respective outputs;
I used a LM324, just because it was on the bench, with a 100nF capacitor between supply pins (4 and 11); any TL074 or similar chip should work the same.
Since the output impedance of the voltage followers is very low, the resistors networks connected to each input of the LM1496 are now electrically symmetrical because of the 180k resistors that I added.
Seen from the LM1496's point of view, it is the same as if the DC inputs were always used (before modification), but this is balanced by the new 180k resistors going to GND.
Trimming is the same as before the modification, I mean the same instructions are still valid.
While I was at it, I also added output offset compensation trimmers to the last opamp of each modulator (U2b and U4b); the trimmers I used are 100k multiturn, connected between power supply rails, and a 2M2 resistor goes from the wiper of the trimmer to pin 6 of the TL072. I am not sure that it actually was worth it, but it's done.
All of this was tested last night, and it works fine: after trimming, no more offsets
The AC coupled input jack and the DC input jack for the same input should not be used at the same time, so I thought of two options:
1- remove the DC inputs and replace them with switches bypassing the capacitor
2- replace the input voltage followers with real active adders; so
DC and AC inputs can all be used together
cheers
Jaba |
|
Back to top
|
|
|
Sebo
Joined: Apr 27, 2007 Posts: 564 Location: Argentina
|
Posted: Tue Jun 21, 2011 7:38 am Post subject:
|
|
|
Hi Jaba, thank you for the mods.
I will try them soon. _________________ Sebo
---------------------------------------
My Music:
https://www.facebook.com/cosaquitos/ |
|
Back to top
|
|
|
yusynth
Joined: Nov 24, 2005 Posts: 1314 Location: France
|
Posted: Tue Jun 21, 2011 11:25 pm Post subject:
|
|
|
Thanks Jaba, I'll try that out ASAP. _________________ Yves |
|
Back to top
|
|
|
Sebo
Joined: Apr 27, 2007 Posts: 564 Location: Argentina
|
Posted: Fri Dec 02, 2011 8:06 pm Post subject:
|
|
|
After some time (almost a year, jeje) I did the mods Jaba suggested (except for the output offset compensators), and they worked great.
Noe the DC inputs works as they should, and I can use the module as two extra VCAs if I want.
Thank you Jaba and Yves. _________________ Sebo
---------------------------------------
My Music:
https://www.facebook.com/cosaquitos/ |
|
Back to top
|
|
|
Benjamin AM
Joined: Nov 04, 2010 Posts: 83 Location: Boise
|
Posted: Sat Dec 03, 2011 10:19 am Post subject:
|
|
|
I was curious if you replaced R37, R15 with the 180k resistor or just wired it in parallel. If the latter is true, would these resistors be necessary? Afterall, the modified resistance to ground(at pin 1) would be something very close to the original value . Last edited by Benjamin AM on Sat Dec 03, 2011 5:03 pm; edited 1 time in total |
|
Back to top
|
|
|
Sebo
Joined: Apr 27, 2007 Posts: 564 Location: Argentina
|
Posted: Sat Dec 03, 2011 4:00 pm Post subject:
|
|
|
I put them in parallel, I though the same, but it just was 2 resistors and a few solder points...
And, to be honest, I don't fully understand why should be added the four 180K resistors, but it works. _________________ Sebo
---------------------------------------
My Music:
https://www.facebook.com/cosaquitos/ |
|
Back to top
|
|
|
Jaba
Joined: Feb 27, 2009 Posts: 48 Location: Genova, Italy
|
Posted: Mon Dec 05, 2011 3:38 am Post subject:
|
|
|
Hi Sebo and Benjamin AM,
here follows my explanation for those 180k resistors:
put them in parallel with the existing ones
The main idea is to "symmetrise" the resistors networks connected to the inputs of the LM1496, after adding the voltage followers that can be seen as the unbalancing cause.
The output of a voltage follower is a very low impedance, so let's think of it as a short circuit to ground; if you add a 180k resistor from pin 1 of LM1496 to ground, then the network made of R10, R12, R13, A2, R14, R15 and the new 180k is symmetrical if you look at it from pins 1 and 4 of the LM1496.
The same applies to the resistors network R3, R4, R5, R6, R7, R8, R9 and the added 180k from pin 8 to ground: it is symmetrical if seen from pins 10 and 8.
Benjamin AM is right, the total resistance to ground does not change much and you can get the same result by replacing R15 with 2660 ohm, or you may even be lucky and reach a perfect balance just with the trimmers and the components tolerances (it did not work for mine). I thought it was easier to look for symmetry and solder an extra resistor here and there than to compute new values and replace existing resistors.
This is just what I did and why; it worked fine for me, so I wanted to share it, because I learnt a lot from this forum and I would like to give my contribution when I can. Seeing that someone actually uses my suggestions really makes me happy, thank you.
I actually ended up implementing real active adders (option 2 at the end of my last message) and I recommend it because it actually gives you more freedom allowing use of DC and AC inputs together.
cheers
Paolo |
|
Back to top
|
|
|
Sebo
Joined: Apr 27, 2007 Posts: 564 Location: Argentina
|
Posted: Mon Dec 05, 2011 4:02 pm Post subject:
|
|
|
Thanks for the explanation Paolo!!! _________________ Sebo
---------------------------------------
My Music:
https://www.facebook.com/cosaquitos/ |
|
Back to top
|
|
|
Benjamin AM
Joined: Nov 04, 2010 Posts: 83 Location: Boise
|
Posted: Mon Jan 23, 2012 1:45 am Post subject:
|
|
|
Quote: | Jaba wrote:
I actually ended up implementing real active adders (option 2 at the end of my last message) and I recommend it because it actually gives you more freedom allowing use of DC and AC inputs together.
|
Then could it be correct to assume that the voltage followers at the inputs would not be needed if one wanted to use this method? |
|
Back to top
|
|
|
Jaba
Joined: Feb 27, 2009 Posts: 48 Location: Genova, Italy
|
Posted: Mon Jan 23, 2012 6:03 am Post subject:
|
|
|
Hi Benjamin AM,
I am not sure that I understand what you mean.
The active adders that I used are the usual OpAmp circuit, so they have a fairly low output impedance and their input impedance can be made high (I used 100k).
The voltage followers were OpAmps too (actually the same ones).
If you put the active adders at the inputs, then the voltage followers are no longer a need; they actually become almost useless, you can forget about them and live happy. I did.
What I actually did is re-use the same OpAmps that were empoyed as voltage followers, rewired to work as simple inverting adders with some 100k resistors. Since -A*-B=AB, you do not need to invert the signals again, they come out right already.
I hope this is an answer to what you asked. If not, please tell me, I'll try again.
Paolo(Jaba) |
|
Back to top
|
|
|
Benjamin AM
Joined: Nov 04, 2010 Posts: 83 Location: Boise
|
Posted: Mon Jan 23, 2012 8:35 am Post subject:
|
|
|
Jaba, That was exactly what I was getting at. Thank you for clarifying.
-Benjamin |
|
Back to top
|
|
|
vcfool
Joined: Jun 19, 2008 Posts: 23 Location: Barcelona
|
Posted: Thu Mar 01, 2012 9:40 am Post subject:
|
|
|
Thank you so much for this!
I have been using the YuSynth Modulator for some time now, and when using the DC inputs I've always suspected that something was wrong. I did experience the DC offsets. Not being and electronics expert I really didn't know what was happening or if that was normal in the design.
Today I spotted this topic and I just did the mod. Seems to work fine, offsets are gone!
I just omitted the 100K resistor to GND on the DC inputs because my modulator doesn't have both inputs (AC/DC) per channel (X/Y), but one input and a switch to bypass the AC cap. I am not sure how to place the resistor only on the DC inputs without making too many changes. Is this resistor crucial? It seems that it is working fine without it.
I just used a TL072 to buffer the inputs and placed the 2 180K resistors (My module is only one Modulator, not dual).
I'm going to modulate something right away and give it a proper test
Thanks again! |
|
Back to top
|
|
|
|