voice modulator. A little advice please.

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hello all

It's seems like forever since I last popped up.

I am in the middle of building a new voice modulator. Those of you who remember me may well remember the last one I made. It's still working great and the members of the projectdalek forum enjoy the free 'how to build your own voice mod' manual.

I have moved on a touch with the design, thanks to Bertus pointing me at an interesting 'ring modulator' circuit.

It is based around the XR2206 IC and has only two other IC's and a handful of components to do it's job.

I have made some changes to the design. The simplest being the addition of a 10K resistor to take power to the input for electret mics. The second change was to changer the pair of LM741 IC's it used to a single LM358.

I'm glad to say the modulator part of the circuit works, as does the sound to light sub circuit, again Bertus had a guiding hand in fine tuning this, and the bridged TDA2003 amplifier.

My first prototype consisted of the modulation circuit and the sound to light circuit on a single PCB. The amplifier is a separate PCB. With them wired to the power supply and using a 10K Log pot to link the audio leads together.

This works very well indeed. Now the thing is it is possible to add an extra pot to the circuit to change the symmetry of the waveform. Also with a switch it is possible to change the waveform from Sine to Triangle. Both of these can have their symmetry changed by the pot previously mentioned.

The waveforms go from symmetrical to saw tooth in either direction. The crux of my question, if you haven't fallen asleep yet =D , is that the triangle wave, or just using the symmetry adjust engenders feedback.

I had this issue on my first voice mod which used a 7555/4066 IC combo to chop the audio. Bertus, yes it's that man again, showed me how to fit a 3 band equaliser between the audio out of the circuit and the in of the amplifier. This was a great success.

So I have made the same 3 band equaliser to go between the modulator and the amplifier. All these sub circuits are on the same PCB.

I had some initial problems with limited volume and the Bass trim pot cutting the audio off but solved this by removing the 100K resistor, marked red, as shown in the accompanying drawings. But the output at the speaker sounds a little 'muddy'?

If I miss the 3 band eq out and just link directly to the amplifier it sounds fine. So what am I missing with regards to the detrimental effect the eq seems to have.

Again, on the EQ image note the pair of 10uF caps highlighted in blue. The modulator has a 4u7 cap on it's output and the amplifier has a 4u7 cap on it's input so these have directly replaced the 10uF's on the EQ.

Here are the schematics for your convenience. Any help would be greatly appreciated.

EDIT: I have just changed the 4u7 caps for a pair of 10uF. Seems a lot better but still not quite 'clean'.

regards

Fenris
 

Attachments

Last edited:

SgtWookie

Joined Jul 17, 2007
22,230
Hello Fenris!
Yes indeed, it's been awhile. Welcome back!

In the 1st schematic, try removing R12, R14 and C9, then jumper from the output of IC3 (pin 7) to your SK2 output; and replace R11 (1M0) with a 100k resistor.

R12 and R13 have been acting as a voltage divider, reducing the volume of the output from the circuit, and R14 had kept the output centered around 0v. However, they also made the output high impedance. The 3-band EQ needs a lower impedance input. Rather than adding another opamp on the input of the EQ to compensate, I'm just using your IC3 as the buffer instead; and by replacing R11 with 1/10th the value cuts the gain of the opamp - so you wind up with just about the same output level, but at a much lower impedance.

C9 is no longer necessary, as C1, the 10uF cap on the 3-band EQ input, provides the DC isolation that's needed.

On the 3-band EQ, I also suggest changing R1 to 75k. This is due to the nature of the TL07x, TL08x, and LF35x family of opamps. Their common-mode range is limited to +V - 1.5v down to -V + 3v. You're using 12v for +V and ground/0V for -V; so your opamp has a range of ~ 3v to 10.5v, with the middle of the range being 6.75v. By reducing R1 to 75k, you shift the average of the reference voltage to ~6.85V with a 12v supply, which is much closer than the 6v is - and gives you a better chance of avoiding "clipping" of the signal.

I'm curious - what software did you use to create your board layout with?
 

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hi Sgt!

Nice to 'see you'. I hope all is well your side of the pond! I tjought there may be an impedance issue but it's one of those vague areas =D for me...yes another one LOL!

I will give it a bash and see how we go.

The software is PCB Wizard. It is distributed by New Wave Concepts Ltd. In the UK Maplins Stock it.

It is a full schematic and PCB design package suitable for double sided PCB creation. It also exports Gerber files in the industry standard. It has a full stock of parts and you can create your own if required.

On that drawing only 7 components are from the standard stock. All the rest I drew using data sheets as reference. Oh and the software will auto route but I wouldn't trust it to tie my shoe laces =D

Using Linux as my OS I have access to several free PCB design programs but I cut my teeth on this one and it is very intuitive and easy to use. It runs under wine on Ubuntu.

Thatoneguy hi there =) Thank you. The overlapping screws are an eyesore I know. When viewed as artwork or 'standard' in the software all that would be seen is a pair of mounting holes. This is because if I use one of the green pots, they have a square body, I wouldn't be able to fit the closest screw in place hence the second screw position.

Right lets try these modifications =D

regards

Fenris
 

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hello again

Well with the suggested modifications done things are much improved. The audio is cleaner. Would I be correct in saying that R13 is now redundant as well?

The volume control now needs backing off 90degrees otherwise I get clipping/distortion and the volume drops. backed of the 90Degrees the volume is reasonably high. There is also less ambient noise from the speaker during periods of no speaking and the null point adjust, removes any trace of the waveform being audibly present in the audio, now functions even better.

I must confess that the volume pot is a linear pot of 10K with a resistor of 1.5K fitted between the wiper and the terminal that is connected to GND. This makes a 'faux log pot' according to a site I read last night. This is only because I was short of a proper log pot. I used the ratio 6:1 but was a second option of 10:1 If I remember correctly these different ratio's were to suit different impedance matches.

Thatoneguy - Sorry I forgot your second question. It's for distorting the voice mainly to suit full size dalek builds =D

Oh and as an aside as you can see that the PCB can be split by separating the EQ/amp side from the sound to light/modulator side. This is to give people the option to use only the sound to light/modulator with an external amplifier of their choice. This would make for a smaller enclosure. The PCB is 127mm X 73mm as one board. IIRC without the EQ/Amp board the PCB is then 78mm X 73mm.

Here's a youtube link to a short demo of the unit PRE modifications

http://www.youtube.com/watch?v=FGGo_QMnjm0


For anyone who is interested see the attached PDF of the original article if you fancy making the unit for music bending =D
regards

Fenris
 

Attachments

Last edited:

SgtWookie

Joined Jul 17, 2007
22,230
Hello again
Hallo there! ;)

Well with the suggested modifications done things are much improved. The audio is cleaner. Would I be correct in saying that R13 is now redundant as well?
Yes indeed, R12, R12, R14 and C9 can all be replaced by a jumper from IC3 pin 7 to the left side of C1 on the 3-band EQ.

The volume control now needs backing off 90 degrees otherwise I get clipping/distortion and the volume drops. backed fof the 90 Degrees the volume is reasonably high. There is also less ambient noise from the speaker during periods of no speaking and the null point adjust, removes any trace of the waveform being audibly present in the audio, now functions even better.
That's good it is functioning better - I erred in thinking that replacing R11 with a 100k resistor would result in sufficient attenuation. I'm not sure where my brain was at the time. :rolleyes: It did reduce the gain of IC3 from 2 down to ~1.1, but it really needed to be reduced considerably to compensate for the divider being removed.

If you would really like to improve the sound quality, you could replace the LM358 with a better opamp. The LM358 is a very old opamp as well as the 741.

I must confess that the volume pot is a linear pot of 10K with a resistor of 1.5K fitted between the wiper and the terminal that is connected to GND. This makes a 'faux log pot' according to a site I read last night. This is only because I was short of a proper log pot. I used the ratio 6:1 but was a second option of 10:1 If I remember correctly these different ratio's were to suit different impedance matches.
That's OK as long as it works. :)

Thatoneguy - Sorry I forgot your second question. It's for distorting the voice mainly to suit full size dalek builds =D
You should see Fenris' Daleks! :eek: They are really something!



Here's a Dalek Storm teaser trailer - I never heard if the movie went anywhere... it's in production, right?

Oh and as an aside as you can see that the PCB can be split by separating the EQ/amp side from the sound to light/modulator side. This is to give people the option to use only the sound to light/modulator with an external amplifier of their choice. This would make for a smaller enclosure. The PCB is 127mm X 73mm as one board. IIRC without the EQ/Amp board the PCB is then 78mm X 73mm.
OK, then keep in mind you will need to have instructions for configuring the boards for the different applications.

For anyone who is interested see the attached PDF of the original article if you fancy making the unit for music bending =D
regards

Fenris
Good deal! Thanks, Fenris!
 

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hi there =D

I have edited the PCB for the unit. Basically I have added silk screen markers.

* to denote components to be removed

# to denote the change in value

====== indicates the wire link required. An extra pad has been added to fit a wire link neatly from the track of pin 7 to the track of the removed component C9.

These will be listed in the instructions to aid the builder.

I presume I could try taking the value of R11 down a bit more to make the entire rotation of the volume pot functional?

I have the op-amps TL072/TL082 in mind. I just had the LM358 IC's to hand.

Ah yes Dalek Storm =) . Built and owned by my friend Alan. I believe some filming has taken place but have heard little of what is happening at this time. I'll have to PM him and see what's happening.

One of your fellow countrymen is building a Storm. In fact I think he may be making more than one. Whereas the one in the image is timber, MDF and filler. Acrodrome, he describes his location as the "NE Quadrant - North America, is building the parts in a similar manner and then using them to make the fibre glass molds to cast all the parts in fibre glass!

As for the PDF file. All praise to Bertus for it is himself who discovered it quite a while ago. In fact it was so long ago he was unable to remember this =D

regards

Fenris
 

thatoneguy

Joined Feb 19, 2009
6,359
Ok, That is extremely impressive!

I'm not into British shows, but that is an excellent build.

Fenris, did you build that from scratch, is it the "actual" or a replica?
 

SgtWookie

Joined Jul 17, 2007
22,230
The TL071/TL072/TL074 single/dual/quad opamps are lower-noise versions of the TL081/TL082/TL084 opamps.

They're certainly an improvement over the LM358, but they really work best in a dual rail enviornment (both positive and negative supplies) due to the -V + 3v limit I was mentioning earlier.

There are many newer opamps that have much better specifications. Even if they cost a bit more, the price difference is not much compared to the cost of an entire Dalek.

It seems to me that the XR2206 could be eliminated with some effort put into a redesign. I'm not certain how long those will continue to be manufactured; it's one of the very few analog (maybe the only one) function generators still being made.

[eta]
Exar no longer has the XR-2206 on its' website - so it is out of production. Very few places have them anymore.
 
Last edited:

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hi Sgt

Yes I am up against end of supply in the future =( The price can be a bit steep as well from some of the suppliers as well. Of course that will only go up the rarer they get.

My other problem of course is that in the dalek prop environment a 12V battery is the supply of choice. So this has ruled out some circuits I have seen as they required a dual 12V supply.

But I am wondering if the modulation part of the circuit could be designed around using a 6V, or less, dual supply. I am aware that an op-amp circuit can be used to create a 'ghost ground?'. So there may be no reason to not try a split supply for that part of the circuit. Maybe even the sound to light circuit could be run off the split supply as it is only really used to switch the higher voltage.

I have a few other circuits lined up. But 2 of them want a 9V dual supply which in both cases is a pair of PP3's. They also use the CA3080 which Bertus told me some time ago could be replaced by the NTE996 IC as the CA is no longer made.

I also have the schematics for a unit based around the LM137000, quite commonly used for dalek voice boxes. I have a vero board version which I need to prove the building of before releasing it on the dalek forum for others to build. This circuit was donated by a members electrical engineer friend.

I also have another variation based around the LM137000 which uses a wein bridge oscillator to provide the effect.

One other variation I have only just become aware of is 'The Gilbert Cell' sldo known as a four quadrant amplifier......So is a Gilbert cell just an XR2206 without a built in VCO? <----- That is a sweeping generalisation but I hope I convey my limited understanding. My brain keeps passing out when I try to understand =D

What sort of redesign would you suggest Sgt? I can see I could keep the 'wet/dry' mix of the existing design for sure. Frequency control tunes to the lowest to highest range of Hz used in dalek history. Null control? Input level can be kept as well. It's just a change of the core circuit isn't it?

ThatOneGuy - Storm was built by a friend and it took around 5 months IIRC. It is based on a dalek that never made it to the screen. However the concept art for the 'Dark Dimensions Dalek' had been drawn. The story would have been to celebrate the 30th year of Dr Who featuring all the surviving Dr character actors and companions along with various of the Dr's enemies, the Dalek being just one.

A CGI artist on the projectdalek forum created a 3 dimensional version on his computer, based on the concept art, but with his own subtle changes to the style. Alan saw this and wanted to build it. The rest, as they say, is history.

My small part in Storm is his chaser lights. Based on the works of our own Bill Marsden and with input from both himself and Sgt on the set up of the transistor switching to drive the 3 super bright LED's per output. Here's a video of them in situ =)

http://www.youtube.com/watch?v=yXOi41dTXyE&feature=related

regards

Fenris
 

SgtWookie

Joined Jul 17, 2007
22,230
Hi Sgt

Yes I am up against end of supply in the future =( The price can be a bit steep as well from some of the suppliers as well. Of course that will only go up the rarer they get.
Unfortunately, that's exactly right. :(

My other problem of course is that in the dalek prop environment a 12V battery is the supply of choice. So this has ruled out some circuits I have seen as they required a dual 12V supply.
That is not necessarily a problem; it is quite possible to create a negative supply from a positive supply using either capacitive (for low current requirements) or inductive switching supplies. One needs to get the frequency of the switcher high enough so that it won't cause audible distortion.

But I am wondering if the modulation part of the circuit could be designed around using a 6V, or less, dual supply. I am aware that an op-amp circuit can be used to create a 'ghost ground?'. So there may be no reason to not try a split supply for that part of the circuit. Maybe even the sound to light circuit could be run off the split supply as it is only really used to switch the higher voltage.
The term is "Virtual Ground" - although over that side of the pond they might call it "Virtual Earth". ;)

I've been looking for some inexpensive four-quadrant multipliers, but I've been discovering that you can't say inexpensive and four-quadrant multipliers in one sentence as they don't really seem to go together. The AD633/AD633A seems to be one of the few remaining, and their price starts at $9/ea from US suppliers.

I have a few other circuits lined up. But 2 of them want a 9V dual supply which in both cases is a pair of PP3's. They also use the CA3080 which Bertus told me some time ago could be replaced by the NTE996 IC as the CA is no longer made.
Perhaps you could provide some links to the circuits you have under consideration?

There are OTA's still being manufactured by other companies.


I also have the schematics for a unit based around the LM137000, quite commonly used for dalek voice boxes. I have a vero board version which I need to prove the building of before releasing it on the dalek forum for others to build. This circuit was donated by a members electrical engineer friend.
Actually, I had been looking at the LM13700 as a candidate; even located a SPICE model for it a bit ago.

I also have another variation based around the LM137000 which uses a wein bridge oscillator to provide the effect.
Links, please! :)

One other variation I have only just become aware of is 'The Gilbert Cell' sldo known as a four quadrant amplifier......So is a Gilbert cell just an XR2206 without a built in VCO? <----- That is a sweeping generalisation but I hope I convey my limited understanding. My brain keeps passing out when I try to understand =D
Gilbert cells can be used in the construction of 4-quadrant multipliers as a building block. I'd kind of wanted to stay away from building the circuit from discrete components like that, as it would make the board considerably larger, would require a good bit more soldering, and there is a lot more to have to troubleshoot when things go awry.

Have a look at how ring modulators were made before ICs:
http://en.wikipedia.org/wiki/Ring_modulator
If push comes to shove, might have to go with something like that - but hopefully it will be avoided. People will have no end of trouble trying to figure out the transformers.

What sort of redesign would you suggest Sgt? I can see I could keep the 'wet/dry' mix of the existing design for sure. Frequency control tunes to the lowest to highest range of Hz used in dalek history. Null control? Input level can be kept as well. It's just a change of the core circuit isn't it?
Not sure what you mean by "wet/dry"?
The original Daleks used 30Hz in the ring modulators. I didn't know that they changed the frequency later.
What I'm thinking of is replacing the XR2206 with something else, and adapting the replacement to the rest of it. The XR2209 is an interesting IC, fairly inexpensive at ~$2.40/ea or less, and it's in production. It doesn't have all of the functions of a XR2206, but I haven't looked at it too hard yet.

ThatOneGuy - Storm was built by a friend and it took around 5 months IIRC. It is based on a dalek that never made it to the screen. However the concept art for the 'Dark Dimensions Dalek' had been drawn. The story would have been to celebrate the 30th year of Dr Who featuring all the surviving Dr character actors and companions along with various of the Dr's enemies, the Dalek being just one.

A CGI artist on the projectdalek forum created a 3 dimensional version on his computer, based on the concept art, but with his own subtle changes to the style. Alan saw this and wanted to build it. The rest, as they say, is history.

My small part in Storm is his chaser lights. Based on the works of our own Bill Marsden and with input from both himself and Sgt on the set up of the transistor switching to drive the 3 super bright LED's per output. Here's a video of them in situ =)

http://www.youtube.com/watch?v=yXOi41dTXyE&feature=related
Gee, I vaguely remember that thread - it was a good while ago.
It really looks awesome on the Dalek Storm! I'll bet that the rest of the Dalek builders now want them too. :)
 

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hi there

I am relieved that the Gilbert cell is ruled out =D

The AD633. Yep I have a schematic for that. Attached

And here are a lot more =D

The traditional ring mod (transformers and diodes) I drew up as a modular design on vero board. Primarily this was to make it as easy as possible for those on a budget to have a go. A few people have made this or used the transformer/diode unit to great effect.

In fact one of my fellow members took his set up to a Sci-Fi Convention and Nicolas Briggs, current voice of the daleks, had a go. He couldn't believe how simple and effective it was. He uses the mooger fooger so I bet he couldn't believe how much that cost him only to have a go on a sub £10 circuit =D

The HE Ring mod and Unimod use the CA3080 and need a dual rail.

Wein Bridge - This is the circuit used in the Traditional unit. On my module The pair of resistors are a double pot so I can adjust the frequency range.

Purves component side - This is the 'to be built for checking circuit' donated by Harry Purves. There is a second small circuit that is used to 'tune' the frequency circuit. It is a fixed frequency. The schematics are on paper not my PC. It uses a pair of TL072's, the LM13700 and an LM380 amplifier. The sound to light is transistor based with a pair of mosfets to handle the load.

LM13700 - Based on a layout of another voice mod but with different sound to light and a solid state wein bridge replacing the original transistor/bulb unit. It also uses the TDA2003 Bridged amplifier. A popular choice =) .The 5V regulator can be dropped, Bertus told me ;).

Bertus also showed us what to do in the event of non CT transformers.

Wet/Dry control - You can adjust the amount of modulation/effect added to the voice. From full modulation to all normal voice and everywhere in between.

The XR2209 seems to be for making the waveforms, though not sine wave =( , It's a shame about the 2206. Very versatile and fun to play with.

regards

Fenris

PS. Just run into the 6 item upload limit. See next post =D
 

Attachments

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hello again :D

I have dug up a couple of schematics you may be interested in. Both are labelled as 'four-quadrant multipliers'.

The first uses the defunct CA3080E, which the NTE996 replaces, The second uses the LM13700.

In both cases I guess a wein bridge could be used and a TL071/72 can be used as a pre-amp if an electret mic is to be used.

Dual supply again.

regards

Fenris
 

Attachments

SgtWookie

Joined Jul 17, 2007
22,230
The LM13700 seems to be the most appealing so far, as they are relatively inexpensive, and seem to require a fairly small amount of parts besides the IC itself.

NTE parts should only be used out of absolute desperation, as they are terribly expensive compared to the original part.

If you have not gone into mass production with your existing boards, then it would make good sense to change to modern parts that will be available for the foreseeable future.
 

SgtWookie

Joined Jul 17, 2007
22,230
One very helpful thing you could do is to make a couple of monaural .wav samples of "exterminate " one with no effects, and one with ring modulated effects, at 8kHz and 16kHz sampling rates. Just a couple of seconds would be adequate.

LTSpice has the ability to read .wav files, process them through a circuit, and output a new .wav file so that you can get an idea of what the processed sound will be.
 

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hi Sgt

You make a good point. I have actually ordered a couple of CA3080E IC's. But these will be to purely entertain myself :D

I have also ordered a couple of the readily available and inexpensive LM13700 IC's to.... Bite the bullet and change the current design which uses the XR2206. I will complete the pair I have, the full version and the Modulator side, by casing them and these will become 'my voice mods'.

Can you believe I build them for everyone else and I haven't got one for myself :D

So I will look at the current PCB with a view to replacing the XR2206. Hopefully there will be enough room for the LM13700, frequency generator (with a view to more than one waveform).

I need a new breadboard. My current one is occupied by a digital clock which I am loath to take down just yet :D

The LTSpice sounds interesting. I know it runs on Linux. During development of the original version of the voice mod I had multisim on windows and was able to directly 'speak' into the virtual circuit. It took a while for it to process the result but I was impressed with it's interpretation.

regards

Fenris
 

Thread Starter

Fenris

Joined Oct 21, 2007
288
Hi there

I have been looking at boost converters and the like and have come across this circuit. It boosts 12VDC to 24VDC @ 800mA. Is it feasible to then add a resistor divider with complimentary caps or an op-amp based circuit to split the 24V into +12V, Gnd, -12.

http://www.circuitstoday.com/12v-to-24v-dc-dc-converter

I appreciate there may be losses but would there also be an adverse effect on how the 800mA is split?

Of course if anyone can point me at a relatively simple positive to negative circuit with enough oomph to drive a few op-amps AND a bridged TDA2003. The TDA is not a must have but is of interest.

regards

Fenris
 
Top