| Author |
Message |
jhaible
Joined: May 25, 2007
Posts: 2014
Location: Germany
Audio files: 24
|
 Posted: Tue Jul 06, 2010 5:03 am Post subject:
|
  |
|
| /mr wrote: | | janvanvolt wrote: | | i'll already start ordering the capacitors (5% will be). What's the ideal value to look for of those two ? |
Go upwards in this thread and check jhaible's posts for the answer.
It's not that far.  |
Thanks. It's also in the component overlay PDF I just posted. Zoom in and see every detail. The 1% caps are marked as such. Only 47nF and 10nF values.
JH.
_________________
"I tell you the truth, if anyone says to this mountain, 'Go, throw yourself into the sea,' and does not doubt in his heart but believes that what he says will happen, it will be done for him. Therefore I tell you, whatever you ask for in prayer, believe that you have received it, and it will be yours." (Mk 11,23f) |
|
|
Back to top
|
|
 |
/mr
Joined: Aug 05, 2007
Posts: 223
Location: Elektron City, Sweden
Audio files: 1
|
| Posted: Tue Jul 06, 2010 5:32 am Post subject:
|
|
|
Are there any good advice about selecting the right capacitor types on a long-term basis? Even if the caps measure within 1% today, they might be out of tune in 20 years. Any rules of thumb when it comes to dielectric materials etc?
Longevity is important. The famous old vocoders mentioned as models for the Living Vocoder are now over 30 years old.
(And they are probably trimmable to compensate some of the aging errors... the Living Vocoder on the other hand relies on high-precision capacitors instead of trimmers...)  |
|
|
Back to top
|
|
|
jhaible
Joined: May 25, 2007
Posts: 2014
Location: Germany
Audio files: 24
|
| Posted: Tue Jul 06, 2010 6:54 am Post subject:
|
|
|
| /mr wrote: | Are there any good advice about selecting the right capacitor types on a long-term basis? Even if the caps measure within 1% today, they might be out of tune in 20 years. Any rules of thumb when it comes to dielectric materials etc?
Longevity is important. The famous old vocoders mentioned as models for the Living Vocoder are now over 30 years old.
(And they are probably trimmable to compensate some of the aging errors... the Living Vocoder on the other hand relies on high-precision capacitors instead of trimmers...) |
Don't worry too much.
Even asuming your capacitance meter is 10% off, you can still use it to select the capacitors for the vocoder with it, because all you need is *relative* precision.
It's not exactly what you were asking, but I'd assume ageing would also go "into the same direction" if you're using capacitors of the same type.
For comparison: There is *no* trimming of center frequencies in the EMS vocoders (not the 2000, not the 3000 and not even the 5000). I don't know what capacitor tolerances they used, but they apparently used *resistors* of 2%!
Synton selected capacitors in 10 groups for their biggest Vocoder model. Probably resulting in 1% tolerance after selection.
Sennheiser made a strange choice of filter topology that is very sensitive to component tolerances - and they consequently had to fine-tune every filter stage by paralleling fixed resistor values with select-on-test resistors. (Probably using a resistor decade)
ETI/Powertran/Elrad Vocoder (from the same designer as EMS) had lower-order filters (4pole vs 6pole in the EMS2000), and required a trimpot per filter for fine-tuning. (You buy lower pole-count with higher component sensitivity for a given filter selectivity.)
So much for vocoder trivia today.
For the filters, I go the EMS 5000 route, without knowing what component values they used (I made my own calculations), but presumably with better component specs (1% vs. 2% or worse). I hope I can't go wrong that way.
JH.
_________________
"I tell you the truth, if anyone says to this mountain, 'Go, throw yourself into the sea,' and does not doubt in his heart but believes that what he says will happen, it will be done for him. Therefore I tell you, whatever you ask for in prayer, believe that you have received it, and it will be yours." (Mk 11,23f) |
|
|
Back to top
|
|
|
TekniK
Joined: Aug 10, 2008
Posts: 1059
|
| Posted: Tue Jul 06, 2010 12:18 pm Post subject:
|
|
|
| jhaible wrote: | So much for vocoder trivia today.
JH. |
You are well informed,its nice and interesting to read |
|
|
Back to top
|
|
|
janvanvolt
Joined: Nov 24, 2005
Posts: 285
Location: Mainz, Germany
|
|
|
Back to top
|
|
|
/mr
Joined: Aug 05, 2007
Posts: 223
Location: Elektron City, Sweden
Audio files: 1
|
| Posted: Thu Jul 08, 2010 4:58 pm Post subject:
|
|
|
| jhaible wrote: | | /mr wrote: | | Even if the caps measure within 1% today, they might be out of tune in 20 years. Any rules of thumb when it comes to dielectric materials etc? |
Don't worry too much.
Even asuming your capacitance meter is 10% off, you can still use it to select the capacitors for the vocoder with it, because all you need is *relative* precision. |
True, glancing over formulas for f and Q in other filters I'd expect that a shift of all capacitor values would just shift the cutoff frequency and leave the Q value unaffected. With less matching capacitors on the other hand, I'd expect mostly the Q value to change enough to become noticeable. [More about this at the bottom of this post.]
| jhaible wrote: | | It's not exactly what you were asking, but I'd assume ageing would also go "into the same direction" if you're using capacitors of the same type. |
Hopefully they would - but still I expect certain cap types to differ more between types, than the individuals differ within the same type. Would be nice to stumble into some old boffin pointing his pipe at me, telling me what dielectric is superior in the long run. 
| jhaible wrote: | | So much for vocoder trivia today. |
Many thanks for that, very nice.
| janvanvolt wrote: | okay, lots of 47 and 10 bought and a capacitor measurement tool.
fun begins on weekend. |
Speaking of capacitor matching and *relative* precision, there are surely some good practice or tactics to follow to make the best out of it.
When I measure and match transistors or capacitors, I use to put a ruler on the table, make up some appropriate scale of numbers, and lay the measured components along the ruler according to their measured values. Then I pick some pairs of really close ones if I'm hunting for matched transistor pairs etc. A similar routine can be used here. Few of the hundreds of transistors will be "the right value", or "the same value", so it would be handy to have them all placed along a long scale and then handpicked. It's as easy as sorting them in some different bins, but gives much more precision.
If Jürgen doesn't directly advice against it, I'd guess that it would be wise to pick out matched groups of 8 of the 47nF caps, start putting the lowest-valued groups in the filters from channel 2, a bit higher in channel 3, then 4, 5 and so on, up to channel 10 (which is the last one with 47 nF caps), and then leaving out the highest-valued 47 nF caps unused(!); then picking out matched groups of 8 of the 10 nF caps, putting the highest-valued groups in band 19, a bit lower in 18, then 17, 16 and down to channel 11, and leave out the lowest-valued 10 nF caps unused.
The choice of lower values for the lower channels and vice versa makes the bands spread a bit wider than their default values.
The omission of the highest-valued 47s and the lowest-valued 10s prevents band 10 and 11 to come too close to each other.
And the overall method suggested avoids having the frequencies and Q values of the bands randomly distributed with some of them closer and some separated, some with sharper peaks and some with broader.
That being said, it could be all wrong and insane, and I'm of course open for all kinds of opinions and corrections. |
|
|
Back to top
|
|
|
jhaible
Joined: May 25, 2007
Posts: 2014
Location: Germany
Audio files: 24
|
| Posted: Fri Jul 09, 2010 1:01 am Post subject:
|
|
|
Interesting idea to minimize the errors!
If you want to find out the final effect on phase cancellation, you can simulate it, if you like. No idea how it will turn out.
What I really recommend is buying the 1% types I've listed on the Living Vocoder web site. Unless you have a *lot* of time, it's worth to have the manufacturer made the selection for you and buy them as 1% already.
Anther idea, if you're set on selecting caps yourself: Use the "better" ones in the synthesis section, and the dropouts in the analysis section. In the latter, it's still advisable to have the 8 caps of one channel matched, but matching from channel to channel isn't as important as in the former.
And finally: Maybe using totally unmatched caps (well, say, still 5% types) might give interesting results also. There would be different level in each channel (when the center freq of partial filters doesn't match the next partial filter) - but you might be able to cure that with the level potentiometers of each channel. There may be phase cancellation fro channel to channel - but that might actually add to the "character" of the sound. Look at the Moog fixed filter banks! - But it would be a random / unpredictable effect. What I really recommend is using 1% caps.
Or measure each cap and correct it with a tiny ceramic cap soldered in parallel on the copper side ...
JH.
_________________
"I tell you the truth, if anyone says to this mountain, 'Go, throw yourself into the sea,' and does not doubt in his heart but believes that what he says will happen, it will be done for him. Therefore I tell you, whatever you ask for in prayer, believe that you have received it, and it will be yours." (Mk 11,23f) |
|
|
Back to top
|
|
|
/mr
Joined: Aug 05, 2007
Posts: 223
Location: Elektron City, Sweden
Audio files: 1
|
| Posted: Fri Jul 09, 2010 1:31 am Post subject:
|
|
|
| jhaible wrote: | | Anther idea, if you're set on selecting caps yourself: Use the "better" ones in the synthesis section, and the dropouts in the analysis section. In the latter, it's still advisable to have the 8 caps of one channel matched, but matching from channel to channel isn't as important as in the former. |
That's an important point as well, and should be used in any selection method.
And it's true that the old vocoders/filterbanks that we like are far from perfect - they have "character". Maybe that's even why we like them. |
|
|
Back to top
|
|
|
jhaible
Joined: May 25, 2007
Posts: 2014
Location: Germany
Audio files: 24
|
| Posted: Sat Jul 10, 2010 10:47 am Post subject:
|
|
|
Any volunteer for creating a BOM?
I know there are always a couple of builders who are faster in making such component lists than I am.
So the question is: Does anybody want to do it officially? I cannot pay you for this, but I'd appreciate your help for me and the community who are going to build this diy-vocoder ...
So if anybody wants to spend a couple of hours going over this: http://www.jhaible.de/vocoder/schematics/living_vocoder_comp_overlay.pdf and creating a component table, please raise your finger!
JH.
_________________
"I tell you the truth, if anyone says to this mountain, 'Go, throw yourself into the sea,' and does not doubt in his heart but believes that what he says will happen, it will be done for him. Therefore I tell you, whatever you ask for in prayer, believe that you have received it, and it will be yours." (Mk 11,23f) |
|
|
Back to top
|
|
|
TekniK
Joined: Aug 10, 2008
Posts: 1059
|
| Posted: Sat Jul 10, 2010 11:42 am Post subject:
|
|
|
you could give a 50% reduction on the PCB for the guy that would do it  |
|
|
Back to top
|
|
|
2thick4uni
Joined: Feb 20, 2009
Posts: 113
Location: UK
|
| Posted: Sat Jul 10, 2010 1:05 pm Post subject:
|
|
|
| TekniK wrote: | | you could give a 50% reduction on the PCB for the guy that would do it |
How does volunteer and unpaid fit in with this? Because of its size alone it will be a very expensive pcb and that would probably represent good pay for a couple of hours work  |
|
|
Back to top
|
|
|
TekniK
Joined: Aug 10, 2008
Posts: 1059
|
| Posted: Sat Jul 10, 2010 2:40 pm Post subject:
|
|
|
| 2thick4uni wrote: | | TekniK wrote: | | you could give a 50% reduction on the PCB for the guy that would do it |
How does volunteer and unpaid fit in with this? |
Well he clearly did wrote he can not pay,its not the same as he don't want to pay.
A discount on a PCB is a sollution here 'if you can not pay'
It was just an idea to find fast a 'volunteer'..
And to make things even more clear,am not 'volunteer' |
|
|
Back to top
|
|
|
jhaible
Joined: May 25, 2007
Posts: 2014
Location: Germany
Audio files: 24
|
| Posted: Sat Jul 10, 2010 3:33 pm Post subject:
|
|
|
Thank you - I already got a PM from a friendly soul who will do it.
JH.
_________________
"I tell you the truth, if anyone says to this mountain, 'Go, throw yourself into the sea,' and does not doubt in his heart but believes that what he says will happen, it will be done for him. Therefore I tell you, whatever you ask for in prayer, believe that you have received it, and it will be yours." (Mk 11,23f) |
|
|
Back to top
|
|
|
jhaible
Joined: May 25, 2007
Posts: 2014
Location: Germany
Audio files: 24
|
| Posted: Wed Jul 14, 2010 4:38 am Post subject:
|
|
|
Preliminary Bill of Materials (BOM)
http://www.jhaible.de/vocoder/living_vocoder_BOM.pdf
A big thank you to Oscar Salas for spending hours to derive this from my componets layout!
Please note that it's a preliminary list. As I'm typing this, the prototype has not even be built yet. (PCB is ordered.)
JH.
_________________
"I tell you the truth, if anyone says to this mountain, 'Go, throw yourself into the sea,' and does not doubt in his heart but believes that what he says will happen, it will be done for him. Therefore I tell you, whatever you ask for in prayer, believe that you have received it, and it will be yours." (Mk 11,23f) |
|
|
Back to top
|
|
|
zthee
Joined: Feb 20, 2008
Posts: 414
Location: Stockholm
|
|
|
Back to top
|
|
|
robotmakers
Joined: Mar 12, 2008
Posts: 7
Location: NY, USA
|
| Posted: Wed Jul 14, 2010 11:26 am Post subject:
|
|
|
Please excuse the out-of-left-field (possibly stupid) question that follows. Given that this project will require what is surely a super-human amount of effort to hand assemble, has any thought been given to the possibility of designing an SMT version of the board, and having a small (pre-paid) run of fully populated boards machine made?
Thanks,
Roger |
|
|
Back to top
|
|
|
loydb
Joined: Feb 04, 2010
Posts: 393
Location: Providence, RI
|
| Posted: Wed Jul 14, 2010 12:11 pm Post subject:
|
|
|
| robotmakers wrote: | Please excuse the out-of-left-field (possibly stupid) question that follows. Given that this project will require what is surely a super-human amount of effort to hand assemble, has any thought been given to the possibility of designing an SMT version of the board, and having a small (pre-paid) run of fully populated boards machine made?
Thanks,
Roger |
This ++
I really wanted to do this, but there's no way I'll do that many SMT components by hand.
I'd buy a partially pre-populated one for sure.
_________________
MFOS Ultimate Expand-o-tron Build Log: http://www.electro-music.com/forum/post-308797.html#308797
MFOS Mini-Controller Build Log [FINISHED!]: http://www.electro-music.com/forum/topic-42968.html |
|
|
Back to top
|
|
|
/mr
Joined: Aug 05, 2007
Posts: 223
Location: Elektron City, Sweden
Audio files: 1
|
| Posted: Wed Jul 14, 2010 3:40 pm Post subject:
|
|
|
| robotmakers wrote: | | ... this project will require what is surely a super-human amount of effort to hand assemble ... |
It surely is a huge project, probably the biggest single DIY project you'll ever encounter unless you try to clone a CS-80 or a big mixing desk — but it's primarily a time-consuming project with lots of similar parts. Mostly labour, not complexity.
Most of the board is comprised of the 20 channels. Many of the individual component values here are either used in all 20 channels, or used a lot within each channel. Over and over again you place a lot of components of the same kind and then you work for some time soldering them.
The very big JH Frequency Shifter project, in contrast, had two much smaller boards with mostly different component values and solutions all over them. Much fewer components in total, but much more "random" and requiring constant attention.
In that sense, I expect the Living Vocoder work to be a bit more straightforward. |
|
|
Back to top
|
|
|
eframp
Joined: Sep 11, 2007
Posts: 25
Location: Atlanta, GA
Audio files: 2
|
| Posted: Wed Jul 14, 2010 7:51 pm Post subject:
|
|
|
I'm up for all the through-hole stuff - the repetition doesn't bother me. it's learning about the quantity of SMT parts that is scaring me off - I know I'm not up for that.
What are the chances of getting a board that has just the SMT parts (it's all bypass caps, right?) pre-soldered, leaving the through-hole parts to be done by hand?
_________________
Eric Frampton, keyboards Atlanta, GA
www.ericframpton.com |
|
|
Back to top
|
|
|
diablojoy
Joined: Sep 07, 2008
Posts: 809
Location: melbourne australia
Audio files: 11
|
| Posted: Wed Jul 14, 2010 8:37 pm Post subject:
|
|
|
while 143 smt caps is a fairly large number to do they really are not that difficult I doubt i could do 143 thru hole components as quickly
if you try i think you will come to prefer them |
|
|
Back to top
|
|
|
eframp
Joined: Sep 11, 2007
Posts: 25
Location: Atlanta, GA
Audio files: 2
|
| Posted: Wed Jul 14, 2010 8:42 pm Post subject:
|
|
|
I need someone to show me the Secret SMT Technique then, 'cause I seem to have a hell of a time with them.
_________________
Eric Frampton, keyboards Atlanta, GA
www.ericframpton.com |
|
|
Back to top
|
|
|
jhaible
Joined: May 25, 2007
Posts: 2014
Location: Germany
Audio files: 24
|
| Posted: Wed Jul 14, 2010 10:08 pm Post subject:
|
|
|
| eframp wrote: | I'm up for all the through-hole stuff - the repetition doesn't bother me. it's learning about the quantity of SMT parts that is scaring me off - I know I'm not up for that.
What are the chances of getting a board that has just the SMT parts (it's all bypass caps, right?) pre-soldered, leaving the through-hole parts to be done by hand? |
If you don't like SMD caps, just solder in old-fashioned ceramic caps with short wires on the bottom side. The pads should be big enough. (I'm using pads for 1210 size SMD caps - these are *huge*!)
But you will find that soldering large (!) SMD parts is *easier* than that!
You probably need only a fraction of the number of bypass caps anyway. I have provided pads for one cap per IC, but that doesn't need you will need them all.
JH.
_________________
"I tell you the truth, if anyone says to this mountain, 'Go, throw yourself into the sea,' and does not doubt in his heart but believes that what he says will happen, it will be done for him. Therefore I tell you, whatever you ask for in prayer, believe that you have received it, and it will be yours." (Mk 11,23f) |
|
|
Back to top
|
|
|
Luka
Joined: Jun 29, 2007
Posts: 1003
Location: Melb.
|
|
|
Back to top
|
|
|
jhaible
Joined: May 25, 2007
Posts: 2014
Location: Germany
Audio files: 24
|
| Posted: Thu Jul 15, 2010 2:14 am Post subject:
|
|
|
| Luka wrote: | | Im more worried about all the wiring than smt, There is about 80 connectors |
But you don't really need most of them.
If you want a Matrix, channel pots, filterbank pots, channel LEDs, you have to connect them somehow. But it's your choice - you can wire what you want, and omit what's not important for you.
JH.
_________________
"I tell you the truth, if anyone says to this mountain, 'Go, throw yourself into the sea,' and does not doubt in his heart but believes that what he says will happen, it will be done for him. Therefore I tell you, whatever you ask for in prayer, believe that you have received it, and it will be yours." (Mk 11,23f) |
|
|
Back to top
|
|
|
Luka
Joined: Jun 29, 2007
Posts: 1003
Location: Melb.
|
|
|
Back to top
|
|
|
|
Forum index » DIY Hardware and Software » Jürgen Haible designs
FAQ
Calendar
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
Chat Room
