>
YAMAHA
STRINGS
SS-30
RACK-MOUNTED WITH MIDI
MIDI STRINGS

Sunday, May 24, 2020

Fun With Filters (Pt.3)

The Might(y) Wah!

Now that I have built the filter module I showed in part 1 of Fun with Filters it's time to take a closer look at it. I decided to call it YAMAWAH and in a later part I will show you th finished article with a printed panel.
You will recall that I sawed off the end of a board from the Yamaha B75 organ and the offcut is the Brass Wah section. Obviously, Yamaha did not provide any details of this filter, other than to say that it's a wah effect for the brass voice of the. Also, I have not been able to work out what filter topolgy the iG02611 VCF chip is, so I could not easily guess what the filter type actually is from the schematic alone. Does it 'wah'?

Blessed Plots


When I was first experimenting with the filter I just noted that it was some sort of low-pass a that it rolled off around 10KHz, but I didn't get any further or measure it.The following is a brief look at the filter response curves comparing with the Roland 521 filter module. I used a white noise source from the Mutable Instruments Kinks module. All the traces are from WavePad and I recorded 5 seconds of audio in each case before plotting the spectrum.

Here's white noise as it comes from my MI Kinks module

White noise from MI Kinks


Here's the same noise through the YAMAWAH with the cut off frequency at maximum - 100%. As you can see there is a major resonant peak. This took me slighty by surprise, but then I realised what was going on. When I was first experimenting I use a spectrum analyser VST in Cubase to check the response. In that case there was no peak evident. Couldn't really hear the peak, although I did note a bit of  'bump' when sweeping through frequencies and it was obvious that something else was going on.

White noise through YAMAWAH VCF set to 100% open.

Contrast that with the same white noise passed through the Roland 521 VCF module in LPF mode with resonance set to 0 and the cut off frequency set to 7/10

White noise through the 521 VCF set to 70%

And now the same but with the resonance set to 5/10

White noise through Roland 512 VCF cutoff 7/10, resonance 5/10


Now setting the YAMAWAH cutoff control to 3 o'clock. There is still resonance and the peak has moved down to around 4KHz

White noise through YAMAWAH cutoff set to 83%


So far, it was much as expected. When the YAMAWAH cut off control is set to 12 o'clock something strange happens though. The resonant peak is around 450Hz, but over at other end, from 4KHz upwards there is an uplift!

White noise through YAMAWAH set to 50% cutoff


Moving back to the 521, a similar cut off frequency setting (4/10) look like this. Which is exactly what you'd expect to see in a low pass filter.

White noise through Roland 521 VCF set to 50% cut off and resonance

Back to the YAMAWAH, and here's the setting at around half past ten.

White noise through YAMAWAH set to 33% cut off

And at fully counter clockwise you can see there's been almost no change in the spectrum. In fact the cut-off frequency only seems to take effect  from half past ten and upwards. This suggests that the control voltage is onlt effective over aroudn 2/3 of the Vee supplied to the iG02611, or about 8V of control voltage.

White noise through YAMAWAH - cut off set to 0%



Finally, let's look at a squqre-wave into the YAMAWAH. This is highly coloured by the 10KHz maxium cutoff frequency and the resonance.


Square wave through YAMAWAH VCF

Learnings

Resonance

The fact that the YAMAWAH is resonating proves that resonace is possible without connecting the C1s and KO/KN pins to C3, which is already fed back to the input. It makes me think that the connection between C1s and KO/KN is responsible for part of the resonance and if I break this connection the resonance will disapear. Adding a variable resistor between these pins could then lead to a resonance control.
I experimented with connecting C3 to KO/KN and seperating KO/KN from each other to no great effect, but now I'd like to return to this again.

First this is what happened when I disconnect the C1s from KO/KN. The resoanace peak disapears


 And then at 50%


 And at 0%


So, it really does seem that this is a notch filter. I suspect that changing the caps will move this around.

Anyway, back to resonance and adjusting the amount of current from the C1s to KN/K0 makes no difference. It's either on or off. Instead I again seperated KN from KO and that too kills the resonance. So, then I added a 100K variable resistor between KN and KO and now I have variable resonance!

Here's the plot with the 100K in line between KN and K0. There's still a bit of resonance compared to open circuit and the cutoff frequency has shifted a bit. 

 
Here is the plot with 50% resonance  - 50Kohm - and the peak is more noticable


And so I now need to add a resonance control!

Here's a video with a demonstration of the resonance control being tested


High Pass?

The second interesting thing about the plots is that almost band-stop behaviour as the filter cut-off moves down. I have been hoping that the iG02610/11 might be suitable for setting up as a high-pass filter as well as a low-pass and that behaviour makes me think that maybe it is possible.
It also reminded me that I had made a modification to the C2 pin. On the B75 the pin only has a capacitor to ground. On the CS-01 that pin is connected to Vee by a 10M resistor. I didn't detect and difference by adding this, but now I'm starting to wonder.

Well, I simply removed this resistor and nothing happened. The up-lift was still present, so whatever is going on it's not that resistor. I will have to start chanaging capacitors if I want to figure this one out!







7 comments :

T-150 said...

so bad you fried the spare chip. still a helpfull investigation, hope you gently procede in studying the chip. dont think you will bother rewiring it the CS-01 way and measuring the responce, but that's what will move the study further.

T-150 said...

so bad you fried the spare chip. still a helpfull investigation, hope you gently procede in studying the chip. dont think you will bother rewiring it the CS-01 way and measuring the responce, but that's what will move the study further.

T-150 said...

- On the B75 the pin only has a capacitor to ground. On the CS-01 that pin is connected to Vee by a 10M resistor.
- in some electones this is even wired the following way: a 4.7M resistance, then the viper of a voltage dividing pot of 100k between -15V and positive GND. in one case the control (likely a trimmer) is named "VCF click adjustment", in the other "cut off adjustment". this definitelly has to be cheked better.

T-150 said...

yet another current thought (and a stupid one). in CS-01 (at least in the manual, i dont have a real thing) the cap at the C1' pin is not 22n but 2n! if it's not a schematic mistake, then probably by this the rising slope of the band rejection is virtually eliminated? anyway, if not a mistake, the difference is 10 times!

in electone schematics i've seen one other example (this definitelly is not common) of unequal caps at C1 pins. they were 1n for C1 and 1.5 for C1'. the cap changing experiment is a need anyway - what the plots would look like?

steve norgate said...

Oh damn, I only just saw these comments :-(
It seems I was not getting notified.
So sorry I didn't reply to your helpful input before T-150.

As you may have seen I kind of gave up and just went with what worked. Hopefully, it will be useful for someone one day though.

Anonymous said...

By the way, if you want a source of these ig02611 chips. The Yamaha PS-3 contains one.

steve norgate said...

Very nice! I'll keep an eye out for any damaged or v.cheap ones