JH Living Vocoder [BUILD THREAD]
- OrganFixer
- Common Wiggler
- Posts: 181
- Joined: Wed Feb 05, 2014 4:38 pm
- Location: New York City
I think jhulk is over there as he quotes things in pence. I think USA was a response to my question.
Regarding the resistor. Is there a
schematic to check against?
Regarding the resistor. Is there a
schematic to check against?
Last edited by OrganFixer on Tue Feb 09, 2016 3:13 pm, edited 1 time in total.
The Living Vocoder is working.
As a summary, I suggest:
Synthesis Input and Compander
Levels:
R2610 (Potentiometer on front panel)
R2612=39K (to add 6dB of extra gain)
R2659 (Switch ExcMode and S_Mode OFF and Ns_Mode ON and adjust the trimmer R2659 for 1.23VRMS at U2606B output. Pin 7) (R2659 could also be on the front panel)
R2645=20K
R2649=7.5K
R2654=10K*
R2655=1.8K*
C2626=47nF*
R2657=20K
R2656=JUMPER
*PCB's Layout says 100K, 18K and 4n7 respectively. Schematics say the suggested values.
Switches:
C2622=100n
C2623=100n
C2624=100n
(These values attenuate “pops” when switching, these are noticeable when unvoiced signal turns on Ns_mode Switch for example.)
Posted Image, might have been reduced in size. Click Image to view fullscreen.
Voiced/Unvoiced Detector
Filter's Cutoff Frequency:
I placed high-pass filter cutoff at 5000Hz and low-pass filter cutoff at 2000Hz.
R2311=33K
R2312=150K
R2309=56K
R2310=24K
Instability Preamp Protection:
Noise or high frequency oscillations in the preamp section will turn ON the unvoiced section.
To achieve protection to this, I considered to use the channel 17 analysis output instead of the high pass-filter, or convert the high-pass filter in a band-pass filter. However I found a more easy solution:
Place a 470pF capacitor in parallel with R2302 and
R2303=JUMPER
Posted Image, might have been reduced in size. Click Image to view fullscreen.
Speech Input Amplifier
I noticed instabilities in this section.
TONE Network:
U2102A's inverting input has long trace and also it is the center leg of the TONE potentiometer. It will be recommendable take special care of this when assembling the Vocoder to the front panel, like to use a short length and coaxial cable (shell grounded).
A tested and working solution is to place a 100pF capacitor parallel to R2120.
An extreme solution could be bypass the tone section. In this case: do not place all components related (Blue labels on the attached picture) and C2111=JUMPER, R2117=20K, TONE 5 pin connector (pin 2 to 3)=JUMPER, R2120=20K with a paralleled 220pF capacitor. This is by now untested.
Preamplifier:
I'm using the chip THAT1510 instead the SSM2019.
My suggested values are R2101=10R, R2102=10R, R2103=1K, R2104=1K, RG0=OPEN, RG1=5R, RG2=Capacitor Electrolytic 10V 3300uF (will cut about 10Hz).
Limiter:
RLIM =6K
RG3=22K
Posted Image, might have been reduced in size. Click Image to view fullscreen.
this is what was recommended for tuning the pcb
and some one asked about the values for resistors on the pcb values if they were correct and the answer was yes apart from the tuning values above
As a summary, I suggest:
Synthesis Input and Compander
Levels:
R2610 (Potentiometer on front panel)
R2612=39K (to add 6dB of extra gain)
R2659 (Switch ExcMode and S_Mode OFF and Ns_Mode ON and adjust the trimmer R2659 for 1.23VRMS at U2606B output. Pin 7) (R2659 could also be on the front panel)
R2645=20K
R2649=7.5K
R2654=10K*
R2655=1.8K*
C2626=47nF*
R2657=20K
R2656=JUMPER
*PCB's Layout says 100K, 18K and 4n7 respectively. Schematics say the suggested values.
Switches:
C2622=100n
C2623=100n
C2624=100n
(These values attenuate “pops” when switching, these are noticeable when unvoiced signal turns on Ns_mode Switch for example.)
Posted Image, might have been reduced in size. Click Image to view fullscreen.
Voiced/Unvoiced Detector
Filter's Cutoff Frequency:
I placed high-pass filter cutoff at 5000Hz and low-pass filter cutoff at 2000Hz.
R2311=33K
R2312=150K
R2309=56K
R2310=24K
Instability Preamp Protection:
Noise or high frequency oscillations in the preamp section will turn ON the unvoiced section.
To achieve protection to this, I considered to use the channel 17 analysis output instead of the high pass-filter, or convert the high-pass filter in a band-pass filter. However I found a more easy solution:
Place a 470pF capacitor in parallel with R2302 and
R2303=JUMPER
Posted Image, might have been reduced in size. Click Image to view fullscreen.
Speech Input Amplifier
I noticed instabilities in this section.
TONE Network:
U2102A's inverting input has long trace and also it is the center leg of the TONE potentiometer. It will be recommendable take special care of this when assembling the Vocoder to the front panel, like to use a short length and coaxial cable (shell grounded).
A tested and working solution is to place a 100pF capacitor parallel to R2120.
An extreme solution could be bypass the tone section. In this case: do not place all components related (Blue labels on the attached picture) and C2111=JUMPER, R2117=20K, TONE 5 pin connector (pin 2 to 3)=JUMPER, R2120=20K with a paralleled 220pF capacitor. This is by now untested.
Preamplifier:
I'm using the chip THAT1510 instead the SSM2019.
My suggested values are R2101=10R, R2102=10R, R2103=1K, R2104=1K, RG0=OPEN, RG1=5R, RG2=Capacitor Electrolytic 10V 3300uF (will cut about 10Hz).
Limiter:
RLIM =6K
RG3=22K
Posted Image, might have been reduced in size. Click Image to view fullscreen.
this is what was recommended for tuning the pcb
and some one asked about the values for resistors on the pcb values if they were correct and the answer was yes apart from the tuning values above
Thanks, that makes things cheaper for me. Definitely count me in. Can I ask for 150 of each value though, as I always like to carry a few spares...jhulk wrote:lektroid im in uk
I saw Sound's reply on there, but the 47k/1k8 issue is not mentioned. It was also noted by Jürgen himself that there was another error on the schematic which was corrected on the PCB. I just don't know which way to turn in this instance...as for the resistor on the other forum all the problems was talked about maybe that info is there but would have to sift through a load of pages to find out
but like you i find the schematics are usually right
I guess I'll leave it out for now, until I have completed the build, then swap them out and see which works best (unless anyone here knows better).
- Synthsense
- Common Wiggler
- Posts: 234
- Joined: Sun Nov 04, 2012 12:40 pm
- Location: VCAlona
I have read the 13700 can be used as a drop in for the 13600, but the 13600 is not forward compatible if the circuit utilises the additional diodes of the 13700.jhulk wrote:my pcb turned up today
are people using the 13600 or 13700
as in his docks on some of his builds he requests 13600 as he tried 13700 and did not work as good as the 13600
I always use 13700 in place of the obsolete 13600 and never found any problems. I'll be using 13700's in mine (socketed). In fact for the money, it's probably a good idea to source 20 of each and swap them out, you'll always use them anyway.
http://randomsource.net/haible/vintagesynthdude wrote:Where are these pcb's available? I have built three different JH pcb's that I got directly from him back in the day and they are pretty much amazing. He was a genius.
There's a few original Haible PCBs in stock.
Cool thanks!LektroiD wrote:http://randomsource.net/haible/vintagesynthdude wrote:Where are these pcb's available? I have built three different JH pcb's that I got directly from him back in the day and they are pretty much amazing. He was a genius.
There's a few original Haible PCBs in stock.
You need to be specific when ordering, I ordered a string filter along with my vocoder board, he only sent one PCB (it needs two identical PCBs, populated differently to work). I had to make a separate order for the other. So make sure you specify 2 PCBs when ordering.jhulk wrote:if you ask him nice there is 1 string filter left
If he's sold out, it might be worth asking if he is likely to do another run; he may have plans on making a RS version, or maybe he will rerun the original. Something to discuss with him...
-
thermionicjunky
- Thyratron Genius
- Posts: 2275
- Joined: Thu May 03, 2007 6:33 pm
- Location: Upstate NY
Jörgen Bergfors claims that the 13700 has a bit less gain. Perhaps that is a contributing factor.LektroiD wrote:I have read the 13700 can be used as a drop in for the 13600, but the 13600 is not forward compatible if the circuit utilises the additional diodes of the 13700.jhulk wrote:my pcb turned up today
are people using the 13600 or 13700
as in his docks on some of his builds he requests 13600 as he tried 13700 and did not work as good as the 13600
I always use 13700 in place of the obsolete 13600 and never found any problems. I'll be using 13700's in mine (socketed). In fact for the money, it's probably a good idea to source 20 of each and swap them out, you'll always use them anyway.
I've just posted about the group buy in the vocoder thread in the SDIY forum on Facebook. Hopefully a few more might be in for them.jhulk wrote:any more for the caps really would like to get 3000 of each or more if possible
How many more people do you need in for the group buy?
I usually pull apart some turned IC sockets for this. Any difference buying them individually?jhulk wrote:what about any other parts
turned pin sockets buy them in the 1000,s makes them a lot cheaper
i only use the gold over bronze turned pin sockets for all chips i find the chip insertion types start to go black over time
done a few memorymoog rebuilds most of the time its those tin plate sockets and headers that go black that make them so hard to tune by changing to turned pin sockets and gold headers and they stay in tune so i make it a matter of course to use the gold types
i never understood why people buy expensive parts then skimp on sockets and cheap ass pots or sliders
if you want something to last do it right and only once
and why i wont buy parts from tayda if you buy in large quantities from mouser or farnell you can get them just as cheap and they usually are not b stock
for the turned pins as inline rows you can get them in 40 row and they have proper cut marks already so they are easy to snip unlike sockets
done a few memorymoog rebuilds most of the time its those tin plate sockets and headers that go black that make them so hard to tune by changing to turned pin sockets and gold headers and they stay in tune so i make it a matter of course to use the gold types
i never understood why people buy expensive parts then skimp on sockets and cheap ass pots or sliders
if you want something to last do it right and only once
and why i wont buy parts from tayda if you buy in large quantities from mouser or farnell you can get them just as cheap and they usually are not b stock
for the turned pins as inline rows you can get them in 40 row and they have proper cut marks already so they are easy to snip unlike sockets

