Author |
Message |
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Thu May 07, 2020 2:50 pm Post subject:
Debugging Steiner VCF build. Help please (solved) |
|
|
Hi all, hope you are all well!
Finished building a Steiner Synthacon filter (using CGS35 board, 1st version) with some mods, please see attached schematics below.
The filter passes audio but no response from Fc pot or CV input.
It seems to be filtering: injecting white noise on all 3 inputs, it sounds like LP or HP, BP being less apparent obv.
Pretty low amplitude at the output though, except when oscillating, which is wild, as expected.
Resonance pot works well, after adjusting my two 10k trimmers.
I'm on 15V
I made slight component changes (see schematics). All is marked in red on the schematics.
The most important change: I used Germanium diodes instead of 1N4148, matching the voltage drop to that of 4148s.
For this, my reference was the "CGS35 Euro" version, with 12 x 4148 versus this pcb, with 6 x 4148. 12 x 4148 should drop 12x0.6=7.2V. I used 26 matched D9V Ge diodes, 26x0.277=7.2V (7+6+6+7).
I assumed taking care of voltage drop being equal than that of 12 x 1N4148 was enough, but maybe here, being used as a resistive element, there is some sort of problem using Ge diodes? Or so many (even that the V drop is the same)?
I used 3 matched caps for the core, reading 2n54 (three beauties from 1965, genet poo variety : )
I used 2N3904 for the core, matched.
Using 2N2222/2N2907 for the Sziklai pair (Hfe 155 and 145. I also selected a 2N5172/2N5138 pair, Hfe ca. 300, but using 2222/2907 for the moment)
Output op amp has 3k9 and 100k pot (plus 330R for safety), to get a gain from aprox 0 to 25, for 2222/2907.
Since I'm just a builder with no real electronics skills, I took good care while building: measured every component before soldering, checked transistor pinout with a tester, matched the 2N3904 and checked every solder point just after soldering with a magnifying glass 😬
Now, before asking for help, I checked the whole circuit point by point following the schematic, Fc pot giving 0 to +15V, all wiring, and all diodes in the core, etc, etc.
I can't find the fault...
Please, help 🙏🏼
.
Description: |
|
Filesize: |
110.4 KB |
Viewed: |
409 Time(s) |
This image has been reduced to fit the page. Click on it to enlarge. |
|
Description: |
|
Filesize: |
2.92 MB |
Viewed: |
394 Time(s) |
This image has been reduced to fit the page. Click on it to enlarge. |
|
Description: |
|
Filesize: |
2.72 MB |
Viewed: |
392 Time(s) |
This image has been reduced to fit the page. Click on it to enlarge. |
|
_________________ Fer
. Last edited by Fernando on Sun May 10, 2020 4:45 pm; edited 1 time in total |
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Fri May 08, 2020 4:50 pm Post subject:
|
|
|
Changes made so far trying to solve it:
- "span" trimmer to 10k (from 2k5)
- control NPNs to BC547 (from 2N3904)
- tried shorting 2 and 4 diodes, one per section.
No change... still CV range is something like 3 semitones, from CV input or Fc pot.
It doesn't go farther even using a 20V p to p sine wave from my generator.
And same (lack of) response no matter if the CV is a 3Vpp or up to 20Vpp wave.
I think I can discard cold joints after many checks.
So, maybe a short somewhere?
or a silly mistake from my confined brain? _________________ Fer
. |
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Sat May 09, 2020 3:27 pm Post subject:
|
|
|
Aaaall right, silly mistake, sorry! I had a 390k in place of a 3k9 under the control transistors.
I meter every component before soldering but it seems that, with this one, my brain agreed that "390.0 k" was correct, oh well
It seems to be working now, I think it should have a bit more of control range, but I'll have to research further to determine if it's working "as it should" or not...
- With TO-18 2n2222/2n2907 it sounds good.
- With old 2n5172/2n5138 (TO-92/TO-106) it sounds amazing!!! (originals in Steiner design)
- My Ge diodes experiment worked!
- I see that the trimmers I added are very important to get the most of this filter.
- It has different resonance sensitivities for each input... (I may take two trimmers to the panel as recessed pots or so)
I also wonder about the effect on sound of the cap's value at the output of the core, before the Sziklai pair. Original and Ken's is 100n, Yusynth 680n, I used 1uF.
I assumed it was a decoupling cap and used something larger but in total ignorance about it's impact and with no tests.
Any light on this very welcome _________________ Fer
. Last edited by Fernando on Sun May 10, 2020 6:24 am; edited 1 time in total |
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Sun May 10, 2020 5:19 am Post subject:
|
|
|
If you refer to the 2n2 caps connected to the diodes, changing their values (which should remain equal) will change where the range of adjustment is. If you increase the size of those caps, the filter will have a lower Fc at a given CV. Smaller caps will raise Fc for a given CV. The quality of the sound will not change. So if the Fc of the filter doesn't go low enough, then a set of larger caps might help. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Sun May 10, 2020 6:27 am Post subject:
|
|
|
Thanks, no.
I mean the cap at the output of the core, from the central cap in the core to the Sziklai pair.
Original uses 100n, Ives use 680n and I used 1uF _________________ Fer
. |
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Sun May 10, 2020 6:58 am Post subject:
|
|
|
I see. Well, that cap looks like it forms a high pass filter. This filter should ideally be designed to have a very low Fc so if the cap was actually too small to start with, increasing it's value might allow more bass, so that could improve the sound. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Sun May 10, 2020 7:13 am Post subject:
|
|
|
that's what I thought and used 1uF...
maybe it was so little (100n) to filter some of the hum? This filter is a bit hummy
and maybe that's why Ives used 680n as a good compromise... _________________ Fer
. |
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Sun May 10, 2020 7:52 am Post subject:
|
|
|
I would rather find the source of the hum rather than change caps to filter it out. The noise is not generated within the VCF, it is coming from outside the VCF, most likely from the PSU or grounding or PSU circuit related problem such as lack of bypass caps. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Sun May 10, 2020 9:54 am Post subject:
|
|
|
>I would rather find the source of the hum rather than change caps to filter it out.
Oh yes. But this filter is generally reported to be hummy. Some shield the core.
So I wondered if such a small cap (100n) is related to it
>such as lack of bypass caps
No bypass caps on the op amp, I'll try that first
Thank you Scott! I thought no humans around here anymore 😅 _________________ Fer
. |
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Mon May 11, 2020 2:19 pm Post subject:
|
|
|
Scott, all,
I'm assuming that decreasing the two 220k resistors at the CV input (one for the freq pot) would result in more CV sensitivity. Am I right?
My plan would be to reduce both to 56k, I have a broader control from my generator, which has greater amplitude (up to 20Vpp), than for instance a Serge Negative Slew (6Vpp max) so I thought that it would be convenient to have greater CV sens.
My plan is also to add a buffer (mixer/processor) to the CV input but let the Fc pot connected directly to one of the CV inputs (220k or lower, 56k)
No idea if 56k will be lowering too much the input impedance of the control transistors
PS: bypass caps added
JovianPyx wrote: | I would rather find the source of the hum rather than change caps to filter it out. The noise is not generated within the VCF, it is coming from outside the VCF, most likely from the PSU or grounding or PSU circuit related problem such as lack of bypass caps. |
_________________ Fer
. |
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Mon May 11, 2020 3:59 pm Post subject:
|
|
|
From what I see, that change should make it about 4 times more sensitive. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Mon May 11, 2020 4:58 pm Post subject:
|
|
|
JovianPyx wrote: | From what I see, that change should make it about 4 times more sensitive. |
Thank you very much, I really appreciate your help
Is correct to consider those two 220k resistors parallel (=110k) and forming a voltage divider with the 1k to ground? (now a 10k trimmer) _________________ Fer
. |
|
Back to top
|
|
|
JovianPyx
Joined: Nov 20, 2007 Posts: 1988 Location: West Red Spot, Jupiter
Audio files: 224
|
Posted: Mon May 11, 2020 5:45 pm Post subject:
|
|
|
Fernando wrote: | JovianPyx wrote: | From what I see, that change should make it about 4 times more sensitive. |
Thank you very much, I really appreciate your help
Is correct to consider those two 220k resistors parallel (=110k) and forming a voltage divider with the 1k to ground? (now a 10k trimmer) |
Yes and no. If the input voltages to both inputs were exactly the same, then it could be considered in parallel. However, both inputs still share the 1K resistor and so then both inputs are part of voltage dividers. The circuit works with currents, so be careful. Your 10K trimmer could allow more current than the converter transistor can handle. That's why the values were set as they were. They were set to attenuate the input by 22 to 1. I'm not an expo wizard, so I can't say at what point you might blow a transistor, but I know that pushing it too hard will eventually let the magic smoke out. So just be careful. _________________ FPGA, dsPIC and Fatman Synth Stuff
Time flies like a banana. Fruit flies when you're having fun. BTW, Do these genes make my ass look fat? corruptio optimi pessima
|
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Mon May 11, 2020 6:13 pm Post subject:
|
|
|
OK, Thanks a lot.
Currently the trimmer has to be set at 750 to 950 ohms to make things work, past that point the whole range is lost. A 1 or 2k trimmer would be enough for fine-tunning.
I'll take good care and try to learn from what you are telling me.
My goal is to have the same degree of Fc control that I get from my Wavetek generator set at about 14Vpp output (= when the filter is being modulated at max), but with 6Vpp signals from other modules. That's why I changed the resistors to 56k, trying to find a way.
Now I see that my generator is controlling the filter with a -7 to +7 signal. And BTOH I read in Dave Brown's page for this filter that negative voltages will take the filter further down in Fc : "You can drive the frequency lower with a negative CV"
So I guess that the greater degree of control using the generator is due to the signal going negative too, not just a larger Vpp signal...
If that's the case, then this filter would expect a bipolar signal to be fully modulated?
JovianPyx wrote: | Fernando wrote: | JovianPyx wrote: | From what I see, that change should make it about 4 times more sensitive. |
Thank you very much, I really appreciate your help
Is correct to consider those two 220k resistors parallel (=110k) and forming a voltage divider with the 1k to ground? (now a 10k trimmer) |
Yes and no. If the input voltages to both inputs were exactly the same, then it could be considered in parallel. However, both inputs still share the 1K resistor and so then both inputs are part of voltage dividers. The circuit works with currents, so be careful. Your 10K trimmer could allow more current than the converter transistor can handle. That's why the values were set as they were. They were set to attenuate the input by 22 to 1. I'm not an expo wizard, so I can't say at what point you might blow a transistor, but I know that pushing it too hard will eventually let the magic smoke out. So just be careful. |
_________________ Fer
. Last edited by Fernando on Tue May 12, 2020 3:22 am; edited 1 time in total |
|
Back to top
|
|
|
gabbagabi
Joined: Nov 29, 2008 Posts: 651 Location: Berlin by n8
Audio files: 23
|
Posted: Mon May 11, 2020 11:39 pm Post subject:
|
|
|
regarding the hum,
when i remember right, some guys have raised the value of that 47µF Caps to reduce hum`n`noize (in cases up to something like 1000µF cos they`ve had nothing else on the bench)
grounding the cases of the potentiometer can also help |
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Tue May 12, 2020 3:21 am Post subject:
|
|
|
gabbagabi wrote: | regarding the hum,
when i remember right, some guys have raised the value of that 47µF Caps to reduce hum`n`noize (in cases up to something like 1000µF cos they`ve had nothing else on the bench)
grounding the cases of the potentiometer can also help |
Thank you. My filter is on the bench now, unshielded (no panel or case) and the hum is low enough. So I will address this later.
I mentioned hum when wondering about the value of the 100n cap. _________________ Fer
. |
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Tue May 19, 2020 5:00 pm Post subject:
a couple of add ons for limiting and timbre variation |
|
|
I love the filter as it is, but while experimenting I found a couple of add ons for limiting and timbre variation. One a decent way of limiting to 10Vpp, the other a nicely raspy clipper.
I tried many Ge diodes and configurations and some zener too.
- Two 5.1V zener in anti-series (K to K) in the feedback loop of the output op amp (inverting in my pcb version), gives a pretty effective and predictable limiter, taming the output to around 10Vpp (otherwise it can go wild to 30Vpp, my unit at least) It can be handy on ocassion.
- One 3.9V zener in parallel with a Ge 1N276 (0.28V@1mA, 0.70V@6mA) from output to ground, gives an interesting timbre, different enough from original signal that I found worth implementing. It clips positive side of the signal in a pleasant way (or negative side if katodes face signal instead of gnd) Pleasant in a Klingon way, don't expect something sweet 🛠 peaks sound like sliding against asphalt.
In this case output is limited to around 6Vpp (you may switch-in a resistor on the op amp to compensate)
I love the filter with it's wild dynamics and beautiful sound (with 2N5172/2N5138 it sounds fierce and liquid) but I think this mods are nice to have on a switch. _________________ Fer
. Last edited by Fernando on Mon May 25, 2020 9:31 am; edited 4 times in total |
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Sat May 23, 2020 5:32 pm Post subject:
final circuit (minus buffers) |
|
|
This is my final circuit, so far :P
Component value changes and additions are shown in red.
After some tests, I'm finally using the original transistors (2N5172/38) for both the core and the Sziklai amp.
Note my pcb is an early one ("rev2" schematic) with inverting op amp, etc.
"CV reject" and "span" trimmers were added (both from Euro version), resonance resistors are now trimmers, etc.
Diodes (26 x D9V), core capacitors, "span" trimmer, "cv reject" trimmer and resonance trimmers, are on a separate board.
I will add CV and audio buffers later.
Description: |
|
Filesize: |
116.42 KB |
Viewed: |
134 Time(s) |
This image has been reduced to fit the page. Click on it to enlarge. |
|
_________________ Fer
. Last edited by Fernando on Fri Jun 12, 2020 12:36 pm; edited 2 times in total |
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Sat Jun 06, 2020 6:18 am Post subject:
final circuit (minus buffers) - last coupling cap |
|
|
(see schematic above)
I'm using a 250k pot and a 10k resistor to get a gain of -25 (I was using 100k pot/3k9 before for same gain)
Note I exchanged positions of the final AC coupling cap and 10k resistor; this way you don't need a huge AC coupling capacitor; I can use a good 1uF cap and have an excellent low freq response.
I see that in CGS35 Euro version Ken did that, exchange cap and resistor positions, I didn't noticed before! I learned the trick a few days ago reading around.
Before, I was using 4.7uF/3k9: the freq response was -3dB at 8.68Hz = "unaltered" response down to 87Hz only, way too high!
Keeping the 3k9 and it's original position (after the cap) I would have changed the cap to 47 or 100uF.
Note for instance the last Yusynth version: Ives use 47uF to go down to 3.39Hz at the output (-3dB 0.34Hz)
Now with a 1uF polypropylene cap I have a great low freq response (and could be smaller)
Useful link to calculate AC coupling cap value, once you know the input impedance: https://www.v-cap.com/coupling-capacitor-calculator.php _________________ Fer
. |
|
Back to top
|
|
|
Fernando
Joined: Dec 30, 2006 Posts: 286 Location: Barcelona & Emporda, Spain
|
Posted: Fri Jun 04, 2021 4:25 am Post subject:
|
|
|
Hi all,
Just as an update: I've been trying many transistors on the Sziklai pair (NPN+PNP output amp). Even some germaniums.
Germaniums in both NPN and PNP positions worked nicely and gave a nice fat sound.
It was key to readjust trimmers when using a new transistor to get the best of it (see my two 10k trimmers under the NPN emitter, above RES pot).
For NPN I had good results with:
Si 2n5172, 2SC2785 (both TO-92, hfe ca. 300)
Ge 2N1304, 2N585.
(I discarded 2N3904, 2N2222, 2N5210, 2N5089)
For PNP I had good results with:
Si 2N5138 blob, 2N5138 TO-18, PN4916, 2N2907A.
Ge 2N1041K (beautiful sound but dimmed highs, fT 0.225MHz), P28, P416b (excellent, fT 80MHz)
A nice sounding pair was 2N5172 or 2SC2785 + P416b
At the end, after literally hundreds of tests, the transistor pair that worked best in all aspects
(freq response, resonance range, SNR, etc) were the original 2N5172 + 2N5138! : )
My 2n5172 is TO-92, hfe 290 and my 2n5138 is an old "blob" type, hfe 260.
. _________________ Fer
. |
|
Back to top
|
|
|
|