Author |
Message |
forbin
Joined: Jan 29, 2009 Posts: 120 Location: Fremantle, Australia
|
Posted: Tue Sep 08, 2009 2:31 am Post subject:
|
|
|
Have a look at this data sheet -- seems better than the one you suggested.
http://www.datasheetcatalog.org/datasheet/vishay/70240.pdf
Q9 - 2N3638A
Q7 & Q8 - 2N3904
Found Q9 was fairly critical as it effected the triangle offset -- I think that I alluded to that in the blog -- the issues that it caused were that the triangle output is offset -- instead of going between -2 and +2 V it went between -1.2V and +3.2V. The three transistors form a sort of Schmidt trigger with a large hysteresis -- this feeds back to drive the OTA and in turn the integrator. I played around with all of this on LTSpice and became reasonably happy with understanding how it all works...
Mark Verbos comments that he wasn't that happy with the design and he was going to take it off his web page. He found that the triangle wasn't symmetrical. The rise and fall times were different. I have checked that pretty carefully in my build and I can't see it. Could be an artifact of the transistor selection?
I think that this is is actually a very clever distilling of the essence of the VCO into something that can be built relatively easily. If you think that it is a bit of pain getting all the FET's and BJT's selected for this then imagine what fun a "perfect" 258 clone would be? |
|
Back to top
|
|
|
ericcoleridge
Joined: Jan 16, 2007 Posts: 889 Location: NYC
|
Posted: Tue Sep 08, 2009 3:04 am Post subject:
|
|
|
forbin wrote: |
Q9 - 2N3638A
Q7 & Q8 - 2N3904
Found Q9 was fairly critical as it effected the triangle offset -- |
Oh, man! I thought it was mainly the FETs that were critical. I'll have to pick up a 3638 now. That's not even listed on the schematic though is it? Did Don suggest that one?
Anyway, I'll be concentrating on getting it to work at all for right now ... |
|
Back to top
|
|
|
zthee
Joined: Feb 20, 2008 Posts: 414 Location: Stockholm
|
|
Back to top
|
|
|
forbin
Joined: Jan 29, 2009 Posts: 120 Location: Fremantle, Australia
|
Posted: Tue Sep 08, 2009 4:11 am Post subject:
|
|
|
Probably not worth a LM394 for that... I seemed to have more grief with Q10 -- I originally had 2N3904's for all the transistors... Probably would work if you tweeked the Bias resistors. I am pretty sure that the SPICE Simulation looked good with all 2n3904's but reality wasn't so kind... Thomas Henry has a very similar design for the VCO1 which get's away with just two transistors. I think that Don used three to generate some of the waveshapes -- Thomas also has extra hi frequency trim components in the log converter... Not sure what transistors Don used originally -- my notes for this are at work. |
|
Back to top
|
|
|
ericcoleridge
Joined: Jan 16, 2007 Posts: 889 Location: NYC
|
Posted: Tue Sep 08, 2009 8:22 am Post subject:
|
|
|
I have a couple questions: the two resistors (R40, R43) that are left out for the Saw version of the 258-- do they get replaced by links, or are they just left out?
Do you recall why you chose the 3638 for Q9? On the schematic, it's a 4248.
I still don't really understand the pin-out for Q10, the 4339. On both datasheets, 'D' is to the left of 'S', and 'G' is to the right of 'S'. Yet, you have them marked opposite this on your PCB. Anyway, as long as it works, I'm not gonna worry about it. |
|
Back to top
|
|
|
forbin
Joined: Jan 29, 2009 Posts: 120 Location: Fremantle, Australia
|
Posted: Tue Sep 08, 2009 5:49 pm Post subject:
|
|
|
The photo is of Q10; the pin labels do seem to agree with the overlay. If you take the power leads as the source end of the FET, the gate is away from the edge and that just leaves the drain...
I think your confusion is that the drain pin is bent across the underside of the package... What i usually do is label the board, label the pins then just make them fit!
Probably best to ignore the Q10 on the right hand inverted board!
I ended up using 2n3638 as i was getting a bit fed up and they happened to be to hand. To a very large extent I build these boards for next to nothing. We have a lot of old stock bits and pieces where I work and very little of what we do these days is through hole technology, it is all surface mount.
The Resistors for the square variant are:
R34 -- open
R40 -- 330k
R43 -- 680k
R44 -- open
R58 -- 2.2k
Don's designs are very clever but you must remember that they are over 40 years old in most cases. A lot of the transistors that we get today although are allegedly the same probably have different manufacturing techniques and are bound to behave a bit differently. FET's in particular have very different gate voltages. The original schematics actually note this, the selection of the Idss must be between .7 and 1.2 -- this is really a huge spread! |
|
Back to top
|
|
|
celadine
Joined: Jul 11, 2006 Posts: 5 Location: unknown dimensions
|
Posted: Tue Sep 08, 2009 6:16 pm Post subject:
|
|
|
Hello,
The FET sine shaper needs some extra trimming for 'perfect' sine, swapping transistors may cause insanity. But, the shaper then probably needs a buffer before it to make this work...and it might screw up the whole waveshape thing...
An example is at bottom of schematic at bottom of this page:
http://home.att.net/~synthsource/ens76/vco2.htm
I would take Mark Verbos's version as an elegant simplification. You won't get perfect sinewaves, but it will still sound great and kick ass, etc. |
|
Back to top
|
|
|
ericcoleridge
Joined: Jan 16, 2007 Posts: 889 Location: NYC
|
Posted: Tue Sep 08, 2009 10:30 pm Post subject:
|
|
|
forbin wrote: |
The Resistors for the square variant are:
R34 -- open
R40 -- 330k
R43 -- 680k
R44 -- open
R58 -- 2.2k
|
Just to clarify, where no resistor is called for-- R34 here, for example-- does one leave this space on the layout empty, or do you replace the resistor with a link?i |
|
Back to top
|
|
|
forbin
Joined: Jan 29, 2009 Posts: 120 Location: Fremantle, Australia
|
Posted: Tue Sep 08, 2009 11:20 pm Post subject:
|
|
|
Leave open (empty)! as Don says infinite...
Oh and I checked during lunch about the transistors. Q9 is a PNP and Q7 & Q8 are NPN's forming a current mirror! I can't have used 2N3904 for all of them! |
|
Back to top
|
|
|
Peake
Joined: Jun 29, 2007 Posts: 1113 Location: Loss Angeles
Audio files: 3
|
Posted: Wed Sep 09, 2009 1:42 pm Post subject:
|
|
|
2N4248s aren't hard to find. And yes, Mark's DIY version has very nearly exactly the same waveshaper section as the original. _________________ We are selling emotions, there are no emotions in a grid. -mwagener
"IC 741. Sometimes you don't want fidelity." -Small Bear Electronics Catalog |
|
Back to top
|
|
|
ultrashock
Joined: Dec 10, 2009 Posts: 40 Location: Vienna.AT
|
Posted: Fri Feb 26, 2010 12:20 am Post subject:
|
|
|
Dear friends!
Could you please explain about the VOLT\OCT TRIM - is it possible to adjust the 258v to 1v\oct response by this trimmer or we have to change whole resistors in the core?
Anyway, how about implementing different v\oct inputs: 1.2 and 1 v\oct simultaneously? For my understating it should not be difficult in recalcucating some resistors.
I'm a big fan of Buchla modules, but let's be realists - the 258v is worth to be set up not only with Buchla modules, but also with all other 1v\oct modulars.
From another side: does 258 verbos adaptation worth to be built in such form as Verbos wrote? For example, we can take the only waveshapers from Verbos' scheme and take vco core from Bergfotron complex VCO (http://hem.bredband.net/bersyn) which has a good 1v\oct response and 1K tempcos (instead of 2k - I have a few 1K smd ones and would like to use them more savingly)- how about this idea?
Could you also explain about Init freq - does it acts as the COARSE pot as in other modulars and Freq acts as FINE (not exactly, but in less range, maybe 1 oct or so)? |
|
Back to top
|
|
|
Peake
Joined: Jun 29, 2007 Posts: 1113 Location: Loss Angeles
Audio files: 3
|
Posted: Sun Feb 28, 2010 5:12 pm Post subject:
|
|
|
IIRC, correct me if I'm wrong, but I believe that Mark's 258 is 1V/octave.
If you use the waveshaping section with another oscillator core, know that it would likely be weird as the Buchla core is triangle, so the waveshaper to sine and square/sawtooth depend upon this. A saw core would produce unexpected results. Those who know more than I do should be posting shortly/hopefully. _________________ We are selling emotions, there are no emotions in a grid. -mwagener
"IC 741. Sometimes you don't want fidelity." -Small Bear Electronics Catalog |
|
Back to top
|
|
|
cbm
Joined: Oct 25, 2005 Posts: 381 Location: San Francisco
|
Posted: Sun Feb 28, 2010 5:45 pm Post subject:
|
|
|
Peake wrote: | IIRC, correct me if I'm wrong, but I believe that Mark's 258 is 1V/octave. |
The Verbos 258v module is .1 v/ semitone (1.2v / octave.) I imagine that it would be fairly easy to convert to .083 v / semitone (1v / octave.) _________________ Chris Muir
http://www.eardrill.com <– My jobby (more than a hobby, less than a job) |
|
Back to top
|
|
|
ultrashock
Joined: Dec 10, 2009 Posts: 40 Location: Vienna.AT
|
Posted: Sun Feb 28, 2010 11:26 pm Post subject:
|
|
|
Peake wrote: | If you use the waveshaping section with another oscillator core, know that it would likely be weird as the Buchla core is triangle, so the waveshaper to sine and square/sawtooth depend upon this. A saw core would produce unexpected results. Those who know more than I do should be posting shortly/hopefully. | that's why I supposed to replace the 1.2V "Verbos" core by 1V\oct Bergfotron Quad Osc core which also is based on the same Buchla one but is more adapted and affordable.
cbm wrote: | Peake wrote: | IIRC, correct me if I'm wrong, but I believe that Mark's 258 is 1V/octave. |
The Verbos 258v module is .1 v/ semitone (1.2v / octave.) I imagine that it would be fairly easy to convert to .083 v / semitone (1v / octave.) |
There is a trimmer (100K, near the ssm2220) which makes v\oct callibration. There are peolpe who built 258v a few times here - could you please answer about this trimmer? (my 258v is on the way to be finished yet dut to the lack of some details, particulary, 3080, but I have plans to retrace this pcb from the beginning in double replacing two 3080 by one accessible 13700). Did anyone manage to do so? |
|
Back to top
|
|
|
ericcoleridge
Joined: Jan 16, 2007 Posts: 889 Location: NYC
|
Posted: Mon Mar 01, 2010 1:03 am Post subject:
|
|
|
The DIY 258 VCO that's been discussed under this topic (posted by Mark on his DIY site) is designed for 1volt per octave scale. It's not the same as the 258v that he sells. |
|
Back to top
|
|
|
cbm
Joined: Oct 25, 2005 Posts: 381 Location: San Francisco
|
Posted: Mon Mar 01, 2010 1:38 am Post subject:
|
|
|
ericcoleridge wrote: | The DIY 258 VCO that's been discussed under this topic (posted by Mark on his DIY site) is designed for 1volt per octave scale. It's not the same as the 258v that he sells. |
Ah... sorry for the confusion. _________________ Chris Muir
http://www.eardrill.com <– My jobby (more than a hobby, less than a job) |
|
Back to top
|
|
|
ultrashock
Joined: Dec 10, 2009 Posts: 40 Location: Vienna.AT
|
Posted: Mon Mar 01, 2010 1:48 am Post subject:
|
|
|
ericcoleridge***
thank you for the explanation! everything is clear now |
|
Back to top
|
|
|
Peake
Joined: Jun 29, 2007 Posts: 1113 Location: Loss Angeles
Audio files: 3
|
Posted: Sat Feb 12, 2011 1:41 pm Post subject:
|
|
|
forbin wrote: | FET's in particular have very different gate voltages. The original schematics actually note this, the selection of the Idss must be between .7 and 1.2 -- this is really a huge spread! | \
Jerk over at the muffwiggler.com is producing a run of Verbos boards and does a tip of the hat to Forbin for these discussions...it is not yet known what is optimal for the waveshaper JFET but indeed, the original design looks for an Idss of right around 1ma...2N3819 have a minimum of 2 and go up to 15 or so!
The 2N4339 isn't a magic fit, however; out of a bag of parts you still have to select some for this characteristic. I look foward to seeing some sort of modern substitute... _________________ We are selling emotions, there are no emotions in a grid. -mwagener
"IC 741. Sometimes you don't want fidelity." -Small Bear Electronics Catalog |
|
Back to top
|
|
|
J3RK
Joined: Jun 05, 2006 Posts: 123 Location: Seattle
|
Posted: Mon Feb 14, 2011 11:09 am Post subject:
|
|
|
I ordered some J201s to try for this one. I've used the 3819 with fairly good results, but I may have just been lucky. I've also tried the MPF102, which I was not as fond of for this purpose.
Also, I'd like to offer these boards here as well, but this is my first production run/sale, so I'd like to do it all through one forum. (since I started this over at Muff's, that's where I'll do it.) The final prototype revs come back within the next few days, and after a good thorough testing, I will start taking orders. If everything starts out well with the ordering process at Muff's, then maybe I will make an announcement for it here as well. I just don't want to get totally overwhelmed the first time out. Want to make sure everyone gets everything correctly and on time.
I will post news here as well going forward, and make my notes available in both places.
I welcome any information beyond what's already in this thread too. Please check out my thread in the DIY Muff forum for project details. I may cross post some of it when I get a minute. |
|
Back to top
|
|
|
Peake
Joined: Jun 29, 2007 Posts: 1113 Location: Loss Angeles
Audio files: 3
|
Posted: Tue Feb 15, 2011 3:33 pm Post subject:
|
|
|
Such a thing as too much of a good thing? Sorry old chap. _________________ We are selling emotions, there are no emotions in a grid. -mwagener
"IC 741. Sometimes you don't want fidelity." -Small Bear Electronics Catalog |
|
Back to top
|
|
|
J3RK
Joined: Jun 05, 2006 Posts: 123 Location: Seattle
|
Posted: Thu Mar 03, 2011 5:24 pm Post subject:
|
|
|
I've started taking orders for my version of this board should anybody want them. I'm going to link the thread on Muff's here. (there are a few reasons for this, but the main one is that I'd prefer not to post my PP address in multiple places if at all possible. The ordering information is in the first post, (and on page 18 ) and there is a ton of information throughout the thread, a couple demos, etc. I'll be posting more demos soon, and all of the information I've compiled from various locations (the main one being this thread.)
http://www.muffwiggler.com/forum/topic-26856.html
Thanks!
I will post my notes, mods list, and other information here as well as soon as I compile it all into something readable. |
|
Back to top
|
|
|
ericcoleridge
Joined: Jan 16, 2007 Posts: 889 Location: NYC
|
Posted: Sat Mar 05, 2011 10:52 pm Post subject:
|
|
|
So, I'm about to power up my second attempt at this board. The first one I blew out-- I think because I had the voltage divider for the "initial frequency" trim wired up incorrectly. When I powered it up, the plastic trim started melting, and I probably blew out all of the chips and transistors...
Is that the correct response when one has the positive and negative voltages wired to the pot lugs in reverse? Or is there some other reason for the blow out?
Now I have negative v wired to lug 1 and positive v to lug 3. |
|
Back to top
|
|
|
synthi
Joined: Jul 20, 2007 Posts: 29 Location: Spain
|
Posted: Sun Mar 06, 2011 8:18 am Post subject:
|
|
|
I think that the only effect because you put the +-V wires in reverse at the pot is that the trim will work in reverse, thats neg to the right and pos to the left. You must have a problem somewhere in the PCB... |
|
Back to top
|
|
|
J3RK
Joined: Jun 05, 2006 Posts: 123 Location: Seattle
|
Posted: Mon Mar 07, 2011 10:23 am Post subject:
|
|
|
Sounds like you may have had a short. That shouldn't happen by connecting the pot in reverse. Could also be a defective pot. I had a pot that got a little piece of wire in it (it was mostly sealed, but had a little slot in the back that it must have gotten in through,) and it caused some problems. (it was more of an intermittent thing though) I've never etched my own boards, but I imagine it would be quite possible for a hard-to-see short to exist. Correct me if I'm wrong. |
|
Back to top
|
|
|
ericcoleridge
Joined: Jan 16, 2007 Posts: 889 Location: NYC
|
Posted: Mon Mar 07, 2011 3:55 pm Post subject:
|
|
|
you are correct.
the thing is, I really prefer etched PCBs now that Im (usually) able to get a pretty clean etch. Besides the pride one can take in knowing that it's hand crafted, particularly for these vintage Buchla, or Arp designs, you get a PCB that's actually much closer to the original (in terms of aesthetics, at least). |
|
Back to top
|
|
|
|